Berkeley Initiative for Soft Computing (BISC)

Fwd: [bisc-group] Re: Are There Cross-Cultural Differences in Reasoning?

2011-05-24T00:54:00.001-07:00

From: <yingxu@ucalgary.ca>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear John and all,

It's a very interesting point of view on the mutual impact between
languages and cognition.

I perceive there are differences in reasoning and semantic cognition when
different languages are used. For example, the key verb in the expression
pattern of English (or Chinese) is in an infix order, i.e., S-V-O and
S-N-V-O where S – subject, V – the verb, O – object, and N – negation.
Therefore, a sentence in English would be such as: "We - do not – speak -
French."

While in Japanese, the express pattern is in a suffix order, i.e., S-O-V
and S-O-V-N. Therefore, the above mentioned example in Japanese would be
such as: "We-French-speak-not." This explains why a listener may not
obtain the key semantics of a sentence in Japanese until the variable
suffixes of the verb are heard at the end of the sentence, especially the
negation suffixes. This uncertainty would makes a listener feels a little
bit nerve in sentence comprehension, which explains why many linguistics
observed that Japanese listeners usually don't smile until the completion
of a sentence – the pending cognition of the "N" suffix may change the
behavior of a language's users.

People also observed that Chinese speakers would usually talk loudly.
There may be social reasons. However, a linguistic explanation for this
behavior is that the minimum energy required to transfer clear oral
information in Chinese is much higher than that of English, particularly
in a noisy environment. This is partially because the tones of Chinese
carry certain semantics, and the highly overloaded syllables (only about
400 x 4 tones) with many different semantics.

In another case, I have an English-speaking colleague who always speaks
loudly. Eventually we found that behavior was cause by a hearing problem.

These three cases seem support that every recurrent thing has a reason
behind.

However, in formal (logical), mathematical, and music languages, the
reasoning processes are the same no matter what one's mother tongue is.

Best regards,

Yingxu Wang
Professor of Cognitive Informatics and Cognitive Computing
Univ. of Calgary, Canada

> I came across an interesting article with the above title.
> See the abstract and conclusion below.
>
> John Sowa
> _______________________________________________________________
>
> http://mentalmodels.princeton.edu/papers/2006xcultural.pdf
>
> Are There Cross-Cultural Differences in Reasoning?
>
> N. Y. Louis Lee
> P. N. Johnson-Laird
>
> Abstract:
>
> Psychologists have suggested that people from different
> cultures use different cognitive processes when they reason.
> Nisbett (2003), for example, proposes that East Asians tend to
> think holistically, dialectically, and on the basis of their
> experience, whereas Westerners tend to think analytically,
> logically, and abstractly. It follows that East Asians should
> tolerate contradictions to a greater degree than Westerners.
> We report an experiment in which East Asians were no more
> likely than Westerners to succumb to illusions of logical
> consistency, and an experiment in which they were no more
> likely than Westerners to reason solely from their experience.
>
> Concluding paragraph:
>
> We leave the last word to John Locke (1690/1959), the
> arch Empiricist whom one might expect to argue that
> individuals in different cultures learn to reason in different
> ways. In fact, he wrote (p. 389):
>
> He that will look into many parts of Asia and
> America will find men reason there perhaps as acutely
> as himself, who yet never heard of a syllogism, nor can
> reduce any one argument to those forms.
> But the mind is not taught to reason by these rules; it
> has a native faculty to perceive the coherence or
> incoherence of its ideas and can range them right
> without any such perplexing repetitions.
>

[bisc-group] What is causality?

2011-05-19T00:35:00.001-07:00

From: Paul P. Wang <ppw@ee.duke.edu>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

I believe the definition of causality itself is 'fuzzy' at best.There are a degree of causality.From extremely precision to very vague.In most cases,the rule/rules is/are not crispy.Nevertheless,it is even better to have some dominant rules [like essential features in pattern recognition] rather than in plain darkness!

Why this issue is important?The success of fuzzy logic depends heavily on this concept!

Leonid's discussion is very interesting,indeed!

Best

Paul

On Thu, 10 Mar 2011, Perlovsky, Leonid Civ USAF AFMC AFRL/RYHE wrote:

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

--
Professor Paul P.Wang
Department of Electrical & Computer Engineering
Pratt School of Engineering
Duke University,Durham ,NC,27708 ,USA

ppw@ee.duke.edu
voice:919 660 5259
fax:919 660 5293

The Theory of Argumentation

2011-05-19T00:29:00.003-07:00

From: Alan Silverman <alanlsilverman@yahoo.com>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

"But people who act on the basis faulty arguments tend to be eliminated from the gene pool."

So, though I may convincingly argue that I can walk on air, if I actually try it (from a 12 story building) I'll never live to pass my genes onto my descendants. What if, instead, I talk someone else into jumping. Then, after he lands, go pick his pockets? That is leveraging risk/reward correctly.

Rules for leveraging risk/reward:

#1: Always leverage risk/reward so that you will gain the greater portion of rewards while others garner the greater risk.

#2. Leverage risk/reward so that rewards accrue to oneself here and now, while risks occur as far away as possible, in time and/or space, and to other people.

#3. In a direct competition between two individuals, the one who chooses short term gains will win over the person who chooses long.

These are winning moves, no matter what the stakes or game. Jared Diamond noted (I believe it was in Natural History) that in primitive society the male who acts responsibly, going out and hunting to provide food for his family, (statistically) ends up the sucker. There's less of a chance that he will pass his genes down than the malingerer who hangs around the tents/huts/apartments, and when the other fellow goes out hunting, sneaks in and bops his wife.

Caveat: So long as society/culture/humanity allow this behavior.

We are discussing the "Truth of Power" versus the "Power of Truth", between yachts and school children.

The Truth of Power says, "Whoever holds power gets to define what is true." In Germany at the conclusion of the 30 Years War, the local prince got to decide whether his territory would become Catholic or Protestant. That is the Truth of Power. The Tea Party today is a manifestation of that truth.

The Power of Truth is objective truth, scientific truth.

The Truth of Power is measured in the length of a human life. It's about competition between individuals in society. In this environment it doesn't really matter what we believe, so long as it serves us in our competition with whomever we compete against.

The Gulf Oil Spill and building a Japanese power plant beside the ocean on a fault line, are glowing examples of leveraging risk/reward correctly. In the first case the people who stand to gain the most from oil drilling don't live anywhere near where the drilling occurs. Or, if they do, they can always afford to move away. In the second, the individuals who championed putting a nuclear power plant in such a risky place have enjoyed the fruits of their victory. Their descendants will not be required to give back their inheritances.

The struggle between the Truth of Power and the Power of Truth has been going on for so long. You may define its beginnings wherever you like. Because of the accelerating pace of technology, of how quickly it can physically change our world, we may finally be approaching the terminus of that struggle.

If individuals who are expert at acquiring wealth and power are the ones least likely to suffer the consequences of their decisions, only the law of unforeseen consequences stands between our world and well deserved disaster.

Best regards,
Alan Silverman

On 5/7/2011 7:25 AM, John F. Sowa wrote:

Dear John, Paul, and Veronica,

JM

Despite your claim (something I do not dispute) that reasoning
may have developed prior to language, it is possible that through
the process of evolution, Mankind's reasoning may have morphed
(or mutated) into a method of thinking in which we no longer
seek the "real" truth, rather we reason to prove our point or
support our decision.

First of all, I make a very sharp distinction between argumentation
and reasoning. Argumentation is a subset of rhetoric. It may use
reasoning, but its purpose is to persuade.

Law suits are classical argumentation. They can be useful in an
evolutionary sense to benefit the speaker. But people who act on
the basis faulty arguments tend to be eliminated from the gene pool.

JM

I guess your statement "I don't believe that at all" kind of provides
evidence for the Theory of Argumentation.

Of course it is. Every email note involves rhetoric. Caveat lector.

PW

... we must always be sensitive (usually quietly so) to the question:
"what definitions are being used/assumed in this assertion?"

I agree. I would replace the word 'reasoning' with the more accurate
word 'rhetoric' in Mercier & Sperber's sentence:

Our hypothesis is that the function of reasoning is argumentative.

VD

I'd hate to think that most of us assign them competitive purposes over
those of general good...

Competition is just one of many uses of rhetoric. And the competitive
version, like any kind of rhetoric, can be used for better or worse.

John

Fwd: [bisc-group] Re: Problem in probability theory/ Snow

2011-05-19T00:29:00.001-07:00

From: John F. Sowa <sowa@bestweb.net>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

On 5/16/2011 11:08 AM, Paul Snow wrote:

I think we divide over the difference between "standard probability
theory has serious limitations" versus "standard probability has its
own proper sphere."

Every technology has limitations.

Sometimes a highly specialized tool that does only one thing very well
is the best choice for that application.

In other cases, a very general tool (such as a Swiss Army Knife) is
useful if you don't have the resources to buy, make, or carry more
than one tool.

Fundamental principle: Anybody who doesn't know the limitations
of a particular tool or technology doesn't really understand it.

John

Fwd: [bisc-group] The Theory of Argumentation

2011-05-07T04:52:00.003-07:00

From: Veronica Dahl <veronica@cs.sfu.ca>
Date: Fri, May 6, 2011 at 10:29 PM

Interesting views.

Which is the "correct" one boils down, temporally and from our
viewpoint, to what purpose us scientists assign to our reasoning
ability and science in general, and to Argumentation Theory in
particular. I'd hate to think that most of us assign them competitive
purposes over those of general good (whichever way one might define
the latter- we can of course dissent here and even err on what will be
of general good, since we never have access to full information from
our little lives' perspective). As a member of Science for Peace, I'd
hate even more the thought that Argumentation's purpose is metaphoric
fighting. In my view, fighting or winning is NOT what life- and
therefore, science- is about. Theory of Argumentation as I understand
it is a tool to help us analyze alternative arguments (say, for or
against different possible medical treatments for a given condition)
in order to determine with as much certainty as possible which
decision might be best. Of course, as the tool it is, we can always
use it for "bad" or simply egotistical ends as well, but it shouldn't
be our purpose.

Best regards,

Veronica

On 05/05/11 6:10 PM, John F. Sowa wrote:
>
> On 5/4/2011 10:30 PM, John Meech wrote:
>>
>> Reasoning was not designed to pursue the truth. Reasoning was designed
>> by evolution to help us win arguments.
>
> I don't believe that at all. Reasoning developed long before language,
> and only a small part of reasoning depends on language. People and
> other animals could never have survived if their reasoning was so
> frivolous.
>
> I would classify arguments as a kind of vicarious fighting by people
> who aren't good with clubs, knives, or swords.
>
> John

Weight could cover both probability and possibility

2011-05-07T04:52:00.001-07:00

From: yoshiki uemura <uemura0742@yahoo.co.jp>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Prof. Zadeh and my member

Soory not to be able to comments of very nice thinking fuzzy sets, as
I deternine to leave Japanese National University. In the next month,
I will visit Warszawa and open semminer entitled "Fuzzy Sets based on
Modeling".

By the way, in risk-analysis, by comparing the expected utility, the
decision maker could select the opitimal decision. The expected
utility can not need for probabity. It need only weight.

As I study only the risk-analysis, perhaps, in the another fuzzy area,
we need only weight. In this area, we not need both probability and
fuzzy.

At last, as I am free, would you please tell me who wont to meet me?
As possible as enjoying the meeting.

Thank you
Best Regards

Yoshiki UEMURA

Fuzzy Sets and Systems

2011-05-03T07:24:00.001-07:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>
Date: Tue, May 3, 2011 at 3:56 AM

Dear Members of the BISC Group:

Sweitze Roffel, Publisher of Physics, Mathematics, Computer
Science & Astronomy, Elsevier, brought to my attention the remarkable
progress and success of Fuzzy Sets and Systems, the premier
international journal in its field. For your information, I am
attaching the data which Sweitze sent to me.

The journal was launched in 1978, with Professor Hans Zimmermann
as its Founding Editor. Currently, it is co-edited by de Baets, Dubois
and Hullermeier. Hans Zimmermann, de Baets, Dubois and Hullermeier
deserve a loud applause. The fuzzy logic community can take pride in
having Fuzzy Sets and Systems as its lead journal. The success of
Fuzzy Sets and Systems owes much to the invaluable support and
encouragement of the highest level administrative structure of
Elsevier.

Regards to all.

Lotfi

--
Lotfi A. Zadeh
Professor in the Graduate School
Director, Berkeley Initiative in Soft Computing (BISC)

Address:
729 Soda Hall #1776
Computer Science Division
Department of Electrical Engineering and Computer Sciences
University of California
Berkeley, CA 94720-1776
zadeh@eecs.berkeley.edu
Tel.(office): (510) 642-4959
Fax (office): (510) 642-1712
Tel.(home): (510) 526-2569
Fax (home): (510) 526-2433
URL: http://www.cs.berkeley.edu/~zadeh/

BISC Homepage URLs
URL: http://zadeh.cs.berkeley.edu/

An american professor reported some IEEE fake conference in March 31, 2011

2011-04-27T02:16:00.003-07:00

An american professor reported some IEEE fake conference in March 31, 2011

from the site: http://dominore.blogspot.com

(also published in http://stopfakeconferences.blogspot.com )

LOW QUALITY CONFERENCES USE THE NAME AND THE LOGO OF IEEE

http://fakeconference.blogspot.com/2011/04/american-professor-reported-some-ieee.html

Re: What is causality?

2011-03-29T07:43:00.001-07:00

Dear Lotfi,

Thank you for your clarification. It ignited a few more questions. In your rigorous shortest path task, a cab driver assumes a known graph and distances. To get this p-valid task he ignores that the road graph may be uncertain. Some roads can be closed, but he does not know this yet. The known graph assumption is a task and context specific decision.

In your f-valid fastest path task, the time for each road is a not-well-defined random variable.

What should be a technique to solve this f-valid task? The name f-valid and your 2009 paper say that fuzzy logic may play a role here. A subjective probability can model or substitute a not-well-defined random variable. A driver may actually use it.

Do we have an argument for a new f-valid technique instead of old subjective probability for this task?

These questions have a more general aspect. Can we justify a specific new technique by pointing out that there is no a universal method to resolve uncertainty and multiplicity of causes. It can justify the need to have a new technique, but not a specific technique that requires much more to be justified.

With best regards

Boris

Boris Kovalerchuk,

Professor, Dept. of Computer Science,

Central Washington University

http://www.cwu.edu/~borisk

>>> On 3/24/2011 at 3:46 PM, in message <4D8BC9B1.1080000@eecs.berkeley.edu>, "Lotfi A. Zadeh" <zadeh@eecs.berkeley.edu> wrote:

Dear Boris,

    Thank you for your constructive comment. Let me clarify what I mean by the taxicab problem. My assumption is that the driver has a street map. More abstractly, what we are dealing with is a graph in which each link is associated with a constant, d, which is the length of the link, and a variable, t, which is the time of traversal. t is a not-well-defined random variable. In version (a), the problem is that of minimizing the sum of d's along a route. For this version, we have shortest-path algorithms, and hence a p-valid solution. For version (b), we do not have algorithms for minimizing the sum of not-well-defined random variables. Thus, for this version we do not have a p-valid solution; what we do have are f-valid solutions. The route chosen by the driver is an f-valid solution. What the taxicab example shows is that there are scientific concepts which do not have p-valid definitions. This applies to the concept of causality, except in special cases in which p-valid definitions of causality may be formulated. So far as I can see, in most real-world settings no existing theory of causality can come up with a p-valid solution. The raincoats problem epitomizes problems with multicausality which are not addressed by the existing theories.

    Regards,

    Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

In memory of a real giant and friend of Fuzzy Logic----Max A.Woodbury

2011-03-22T00:32:00.001-07:00

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

In memory and celebration of the life of Max A.Woodbury
-------------------------------------------------------

Max's wife informed me several months ago the passing of Dr Max
A.Woodbury on January 30,2010.Max is a very well respected and
extremely well known for his research works in probability and
statistics area.He ,so happened ,has been a strong supporter of fuzzy
logic for so many years ,began with,like many other examples,a drink
at a California bar [he told me the story himself] with Lotfi!

He and his associates have had a long tradition of supporting the
JCIS,Joint Conference of Information Sciences [formerly the FT&T,Fuzzy
Theory & Technology Conferences] through out its 15 years history.This
tradition,as well as their papers in Information Sciences Journal by
Elsevier,constituted a corner stone of integrating fuzzy logic into
the main stream of computer sciences.Max's contribution to such
integration can not and should not be under-estimated.We have lost a
true friend who was already a giant in probability and statistics,our
elder brother and sister of the mathematics of uncertainty
[www.ee.duke.edu/math-uncertainty/...,or Goggle or Yahoo!...SMU].

The last decades of his career at Duke University,his reputation and
research efforts were
awarded via a series of Federal grants in his research projects to
such an extend that a stand- alone building housed his associates in
Center for Demographic Studies even after his retirement as a
Professor Emeritus.

For more information,one may find in the web site;

www.tributes.com/show/Max-Woodbury-87865355-cached/....

Sincerely

Paul

--
Professor Paul P.Wang
Department of Electrical & Computer Engineering
Pratt School of Engineering
Duke University,Durham ,NC,27708 ,USA

ppw@ee.duke.edu
voice:919 660 5259
fax:919 660 5293

Fwd: [bisc-group] Earthquake

2011-03-22T00:30:00.001-07:00

---------- Forwarded message ----------
From: Wunsch, Donald C. <dwunsch@mst.edu>
Date: 2011/3/19

Dear Colleagues,

I completely agree with Prof. Huang.

The late Dr. Arlin Cooper, Distinguished Member of Technical Staff at Sandia National Laboratory, was a pioneer in this area. He was fascinated by the potential for fuzzy logic to represent human expert knowledge and by the opportunity to associate membership functions with ranges of reasonable probabilities in the absence of sufficient data sets to do otherwise.

He published some valuable work relating fuzzy logic to risk tree analysis, some of which is referenced in the papers from my previous posting. Much of that work was instigated or inspired by Arlin. He was particularly involved in application of this and other approaches to High Consequence Surety, including nuclear surety and other high consequence scenarios.

I’m sure that, if he were still with us, he would be finding valuable ways to learn from the current catastrophe.

Yours truly,

Don

Donald C. Wunsch II, Ph.D. EE, MBA, PE
Mary K. Finley Missouri Distinguished Professor
Missouri University of Science & Technology
Dept. of Electrical and Computer Engineering
301 W. 16th St, 131 EECH
Rolla MO 65409 USA

wunsch@ieee.org

http://people.mst.edu/faculty/dwunsch_profile.html

http://www.linkedin.com/in/wunsch

(573) 341-4521 Office
(573) 341-4532 Fax

Missouri University of Science & Technology
is known as Missouri S&T for short. It was

previously known as the University of Missouri - Rolla (UMR).

From: hchongfu@bnu.edu.cn [mailto:hchongfu@bnu.edu.cn]
Sent: Thursday, March 17, 2011 10:58 PM
To: zadeh@eecs.berkeley.edu
Cc: bisc-group@lists.eecs.berkeley.edu
Subject: Re:[bisc-group] Earthquake

Dear Prof. Zadeh,

What I have to say is that, it is time for fuzzy community to pay some attection on risk issues. We are in a risk society. Fuzzy logic can help humanity to analyze risks with incomplete information.

The new journal "Journal of Risk Analysis and Crisis Response" (http://www.atlantis-press.com/publications/jracr) will provide a platform for that.

Sincerely
Chongfu Huang, Professor, Ph.D.
President, Society for Risk Analysis - China (SRA-China)
President, Risk Analysis Council of China Association for Disaster Prevention
Address:
Academy of Disaster Reduction and Emergency Management
Beijing Normal University
No.19 Xinjiekouwai Street
Beijing 100875, China
Tel/Fax: +86-10-58805479
E-mail:hchongfu@bnu.edu.cn
------------------ 原始邮件 ------------------
>From: "Lotfi A. Zadeh" <zadeh@eecs.berkeley.edu>
>Reply-To: bisc-group@lists.eecs.berkeley.edu
>To: bisc-group@lists.eecs.berkeley.edu
>Subject: Re:[bisc-group] Earthquake
>Date:Tue, 15 Mar 2011 17:43:37 -0700
>
Dear Members of the BISC Group:

    Professor Kaoru Hirota, President of IFSA, sent me a harrowing account of his experience. I have taken the liberty of posting his account to the BISC Group.

    Regards to all.

    Lotfi

-------- Original Message --------

Subject:	In Big Tokyo Area, we are safe from Magnitude 9.0 Earthquake (message No. 3) (K. Hirota)
Date:	Tue, 15 Mar 2011 15:49:54 +0900
From:	KHirota（�田） <hirota@hrt.dis.titech.ac.jp>
To:	Hirota <hirota@hrt.dis.titech.ac.jp>

Dear all,

   Thank you so much for your inquiry about us by not only e-mails but also international telephone calls. Especially I am very glad to receive a kind message from Prof. Zadeh.

   We (my family, lab members, friends, relatives) are OK. I have already sent the after earthquake reports, cf. message No.1 and No.2 in the below, by BCC to 20+50=70 oversea friends. This is the message No.3 to another 30 friends who sent me inquiry after I sent the below message No.2.

   As I mentioned in the message No.1, I was in the train to the Narita new Tokyo international airport and not in my office. Since my office is far (several hundred km) from the earthquake source, the vibration was not so huge as that in near the source area. But my assistant professor Dr.Ms Fangyan Dong and my secretary Ms Hoshino, who were in my office at 9th floor in 11th stories building, say it was big enough, i.e. almost 1.5 meters width vibrations. Our building itself is strong enough for such big vibration but many things such as books, magazines, ... were fallen down (cf. a photo in my student room). Maybe you noticed that this photo says nothing compared with the photo or the scene of tsunami damaged area uploaded on the internet or TV news.

   Japanese buildings were mostly strong enough for magnitude 9.0 earthquake itself but the protection was not enough against the huge tsunami. I have already visited several times to the tsunami damaged area (several years before this time tsunami), where I saw a heavy 10 m high movable steel gates against tsunami. But this time, it was no use.

   The huge tsunami made big damages to the cooling system motors for the atomic power plant. Currently, as you know, it makes a big problem. Our prime minister Naoto Kan, whose major was applied physics in our university Tokyo Institute of Technology, is now chairing the atomic power plant protection team. We do hope he will surely solve this difficult problem.

   In the big Tokyo area, the planned power outage system (electricity will be stopped for a few hours by area wise rotation to avoid the sudden electricity stopping because the damage of power plant made the shortage of the enough electric power energy supply) has been introduced. My office campus and another campus (where our university network center exists) belong to the different time zone of such planned power outage. So my internet access will be impossible during the both campus power outage periods. That means I maybe not able to access to the e-mail so often from now on.

   That's all for today. With the best wishes.

                                                                                        K. Hirota@office

******message No.2  from Hirota to 20 oversea friends sent Sunday (March 13) 15:30***********

Dear all,

   Thank you for your inquiry about us against the earthquake. The magnitude was revised from 8.8 to 9.0 (world 4th biggest magnitude followed after the third one magnitude 9.1 @ Indian ocean a few years before). On TV news and internet, such miserable scenes of "tsunami" and others have been reported repeatedly. If we think about the strength of this earthquake and the population density in our country, the anti earthquake devices functioned well but we must take care various issues from now on. Anyway, in big Tokyo are, we are safe and currently no big problem.

   Until yesterday afternoon I received about 50 inquiries from many friends. After that until now (Sunday 15:45) I received about 20 more inquiries including yours. So I would like to send you by BCC the below yesterday message from me.

   With the best wishes.

                                                         K. Hirota@home

******message No.1  from Hirota to 50 oversea friends sent Saturday (March 12) 19:30********

Dear all,

   Thank you so much for your inquiry about us against the Japanese

biggest (magnitude 8.8) earthquake yesterday. This e-mail is sent by BCC

to about 50 people from me Kaoru Hirota.

   Yesterday I was expected to fly from Narita airport to Manila by ANA

(All Nippon Air) for conference. Anyway I could come back to my home

after 24 hours "traveling" from my house to Narita New Tokyo

International Airport and coming back. The below is my report about the

earthquake from my viewpoint.

   Now I opened my e-mail box after 24 hours silence and found that

more than 50 inquiries in the last 24 hours about myself or my related

people against the earthquakes.  Anyway now I can send you this e-mail

that means I am OK. And for your, i.e., for more than 50 oversea

friends', reference, I would like to make a little bit long report about

my experience in the last 24 hours in the below.

   The first big vibration happened at 14:48 yesterday when I was

sleeping in the train to Narita airport. I woke up because I felt

intuitively something different atmosphere inside the train which was

stopping inside a dark tunnel. The train was not crowded (around 20

passengers in the same car) but it was so silent except some strange

sounds of something.  I noticed soon that the "something" was earthquake

because I was almost thrown away from the seat even the train was

stopping. It continues for 5 min, then stopped and again vibrating

(repeatedly). Slowly, the train announce informed us that we were

stopping because of a big earthquake, and that inside the car is safer

so please keep stay in the car and not go out because railway tunnel is

very dangerous.

   Several times vibrations, announces, and silence continued for 20

min. After that, the announce said "some officers of our train company

came and they will escort you to the next train station". So all

passengers (more than 100) got off the car and walked inside the tunnel

for 15 min and open air railway path for 5 min to the next railway

station, during this railway walking  several times big vibrations were

added. It (walking the railway) was not easy for most of the people

because they had big luggages for oversea traveling.

   After arriving the railway station the officers there requested us

to go out the train station building because inside is not safe. I

checked my watch and found it was a little bit before 16:00, i.e., about

two hours before my flight departure to Manila by All Nippon Air. At

this point, I (and most of the passengers) did not recognize that the

first big vibration was Japanese biggest (magnitude 8.8!!) earthquake,

the biggest in Japanese long earthquake observation history for more

than 200 years. So I considered how to reach Narita airport from the

railway station to catch my airplane (now I think this was a bad choice,

instead I should make an effort to come back to my house.) All trains

and buses were stopped and only a few taxis were coming occasion by

occasion. Soon several groups were made and shared the same purposes,

i.e., aiming to reach Narita airport smoothly. My group (consisted of 4

members) negotiated with taxi drivers to take us to Narita airport. But

most of them refused because very heavy traffic jam and the express way

to the airport was closed. After several tough negotiations we could

find one taxi driver who agreed. He said, however, I will do my best but

there is a high risk for arriving on time.

   Anyway we started the railway station around 16:15 to the airport

(the distance was about 20 km, note that it is about 100 km from my

house to the airport). The driver requested us to stop for one min when

we arrived at his company base/office along the way to the airport,

where he brought a small portable TV set. By watching the TV, we noticed

that the earthquake was terribly big we have never experienced and so

big damages by tsunamis and strong vibrations in various cities. The

taxi driver was very skilled and we could arrive 5km point to the

airport around 17:30. At that time, we have already known that most of

the airplane departure have been delayed or canceled. So we hope that we

can catch the delayed flights. But the last 5km was terribly crowded and

we could arrive the airport around 19:30, i.e. the taxi speed there was

only 2.5km/hour.

   I rushed inside the airport terminal building and found so many

passengers were given sleeping-bags and small food but I could not

receive them because already run out. I asked several things such as

flight schedule from now on and hotel reservation   to the officers of

All Nippon Air by showing star alliance gold card, and found that such a

card did not make any sense in such occasions. I was informed that all

flights after the earth quake occurred have been canceled, tomorrow

schedule maybe also no good because of the airplane arrangement

problems, most of the hotels near by were closed because of partly

electric power or water down, all public transportation systems

including express way were closed with the exception of local taxis that

can not go back to center of Tokyo.

   The mobile phone and public telephones were almost no use (possible

to connect after 20 times trial in average). All the passengers had

nothing to do except enjoying overnight stay at the airport terminal.

All chairs were occupied and good spaces to sit down were also occupied.

After watching big television screen for 30 min, only I could do was

leaning to the wall until tomorrow without sleeping, eating, and

drinking. During the whole night, I investigated and made a decision not

to visit Manila this time because even if I can fly to Manila tomorrow

the conference will be over when I arrive.

   Very very slowly the morning came. There was no good symptoms.

Around 6:00 this morning, my former student Prof. Kaino @Aoyama Gakuin

University gave me a mobile phone call and he proposed to come here from

his house Kamakura by his car to pick me up to my house. After

considering his health condition (only one year has been passed since

his very heavy surgery operation), the distance about 130 km, all

expressways closed, and very very heavy traffic jam on general roads, I

said to him thank you but no need. Just after that, another proposal

came from my assistant professor Fangyan Don saying that her husband Dr

Chen was starting by his car from his house at Kawasaki to Narita

airport (closer but still 110 km). After that call, i.e., around 6:40,

suddenly the shutter to the railway station opened and the announce said

that private train Keisei sky liner is still closed but JR regular train

is now ready to start (but only to Chiba that locates in between Tokyo

central and Narita airport).

    I was lucky because I happened to be standing very close to the

shutter, and I could get a luxurious first car sheet by paying an extra

charge. When the train started, the regular cars were occupied by 200%

passengers and even the first class car was 130% passengers. The maximum

train speed was limited to only 35 km/h because of the safety reason

after the big earth quake and sometimes stopped here and there, but I

could sleep for 1.5 hours until I arrive at Chiba station around 9:00.

There again the train operation has been stopped but soon the announce

informed the possibility of regular train operation to Tokyo station

soon. Fortunately I could buy again the 1st class seat because there was

a ticket machine just in front of me and enjoy sleeping for another 1.5

hours under the condition of 250% regular car occupation and 150% 1st

class car occupation.

    In JR Tokyo railway station so many passengers were stopping and

moving in various directions. I have never seen such crowdedness before.

Anyway after waiting 1.5 hours there, I could catch Keihin-Tohoku line

train to Ooimachi. It started to moving in that time but it has no 1st

class car and max speed was again limited to 35 km. So it takes for 60

min (usually 20 min) to reach Ooimachi from Tokyo in more than 300%

crowdedness.

   At Ooimachi, the private line Ooimach line was functioning in almost

normal. (Private is far more better than National/Government???) I could

come back to my house within 20 min after the departure from Ooimach

(including walking time from the nearest railway station to my house).

    This is my last 24 hours report.

    Reports of my related places/people;

My first place: Several of my world wide collections (dolls and small

goods) were falling down from the shelves, but practically no problem,

My second place: no problem reported by my sister,

My other places: maybe no problem, I guess, because they are far from

the earthquake source points.

 My office: my assistant professor Fangyang Dong and secretary

Hoshino-san reported that they experienced so slow and big movement

(about 1 m) because our office locates in 9th floor, more bigger

amplified vibration. They have never experienced such big movement

before. So many books and small goods were fallen down. To some extent

both of them recovered but still so many recovery works (that only I can

do) are necessary. Such jobs are waiting for me in the next Monday. But

fortunately no other damages. They said in other labs bigger damages

have been given to PCs and experimental devices. Since all the trains

from Suzukakedai (our campus) were stopped yesterday, Hoshino-san stayed

in our campus with her friend (secretary) and came back to her house in

this morning and found no practical damage. The assistant professor

Fangyan Dong was picked up by her husband's car yesterday late night,

but it took for 8 hours to come back to her house (usually less than one

hour, but this time they arrived in the morning today because of heavy

traffic). Her house was terribly vibrating that was reported by their

kid but practically no problem except for electric power down in that

area. It was very lucky for them because they moved the heavy piano from

the 2nd floor to the 1st floor very recently. Now we have a visiting

scholar, Prof. Pedrycz with his wife @Tokyo Tech Guest House. They said

practically no problem. Maybe they had a very good?? experience because

this was the Japanese (maybe world) biggest earthquake. My students are

also safe.

My relatives and other friends as far as I know by now, no problem. The

reason mainly is because that all of my related people are living in a

little bit far from the earthquake source. But such area in Touhoku

(east north) district including Sendai-city, currently more than 1500

people have been died or unknown. Maybe from now on, the number will

increase.

That's all for today. With the best wishes.

K.Hirota@home

-- ************************************************************ Prof.

Kaoru Hirota Dept. of Computational Intelligence and Systems Science

Interdisciplinary Graduate School of Science and Engineering Tokyo

Institute of Technology G3-49, 4259 Nagatsuta, Midori-ku, Yokohama

226-8502, Japan Tel: +81-45-924-5686/5682 Fax: +81-45-924-5676 e-mail:

hirota@hrt.dis.titech.ac.jp URL: http://www.hrt.dis.titech.ac.jp

*************************************************************

Fwd: [bisc-group] EARTHQUAKE-TSUNAMI

2011-03-22T00:29:00.003-07:00

If you ever want to remove yourself from this mailing list,
you can send mail to <sympa@lists.EECS.Berkeley.EDU> with the following
command in the body of your email message:
unsubscribe bisc-group
or from another account,
unsubscribe bisc-group <your_email_adress>
*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Prof. Huang,

I hope the fuzzy community will take your timely call for paying
attention to risk issues seriously. If I'm not mistaken, the grey
community has been doing so already for several years.

Please allow me to remind you of the highly acclaimed work of Howard
Kunreuther (http://opimweb.wharton.upenn.edu/people/faculty.cfm?id=37),
J. Scott Armstrong (http://www.jscottarmstrong.com) and Ulrich Beck
(http://en.wikipedia.org/wiki/ulrich_beck). They might be willing to
contribute to your new journal.

Respectfully yours,

Hans Kuijper

----- Original Message -----
From: hchongfu@bnu.edu.cn
To: zadeh@eecs.berkeley.edu
Cc: bisc-group@lists.eecs.berkeley.edu
Sent: Friday, March 18, 2011 4:58 AM
Subject: Re:[bisc-group] Earthquake
*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

________________________________

Dear Prof. Zadeh,

What I have to say is that, it is time for fuzzy community to pay some
attection on risk issues. We are in a risk society. Fuzzy logic can
help humanity to analyze risks with incomplete information.

The new journal "Journal of Risk Analysis and Crisis Response"
(http://www.atlantis-press.com/publications/jracr) will provide a
platform for that.

Fwd: [bisc-group] Fwd: Rule-Based Cell Systems Model of Aging using Feedback Loop Motifs Mediated by Stress Responses

2011-03-22T00:29:00.001-07:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>
Date: Tue, Mar 22, 2011 at 12:48 AM
Subject: [bisc-group] Fwd: Rule-Based Cell Systems Model of Aging using Feedback Loop Motifs Mediated by Stress Responses
To: bisc-group@lists.eecs.berkeley.edu

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Members of the BISC Group:

    My good friend, Stephen Coles, Cc me on a message which I thought might be of interest to the BISC Group. I am forwarding the message.

    Regards to all,

    Lotfi

-------- Original Message --------

Subject:	Re: Rule-Based Cell Systems Model of Aging using Feedback Loop Motifs Mediated by Stress Responses
Date:	Mon, 14 Mar 2011 09:21:27 -0700
From:	L. Stephen Coles, M.D., Ph.D. <scoles@grg.org>
To:	Eight Pennies <arthrostylidium@gmail.com>
CC:	James P. Watson, M.D. <jamespwatson@mac.com>, Gerontology Research Group <grg@lists.ucla.edu>, Lotfi A. Zadeh, Ph.D. <zadeh@eecs.berkeley.edu>

Eight: Thank you for this item. My former Professor at UC Berkeley, L. A. Zadeh, would be proud to know
that his invention of Fuzzy-Sets/Fuzzy-Logic was actually being used in developing a Vicious Cycle (VC) theory
of aging at Harvard in Boston and Drexel in Philadelphia. -- Steve Coles

Dr. Coles: FYI - I'm just archiving, but I thought I'd send it to you. Note the futility of a single-point intervention.
Estimate the real-world time-frame to procure a reasonable internal "fix" of our biological systems. -- Eight

L. Stephen Coles, M.D., Ph.D., Co-Founder
Los Angeles Gerontology Research Group
URL: http://www.grg.org
E-mail: scoles@grg.org
E-mail: scoles@ucla.edu

Fwd: [bisc-group] What is causality?

2011-03-18T05:38:00.001-07:00

From: Boris Kovalerchuk <Boris.Kovalerchuk@cwu.edu>
Date: Thu, Mar 17, 2011 at 7:53 AM

Dear Lotfi,

In the raincoat example, you used a common sense setting. In a more scientific setting, this example will be written differently in line with Leonid's and Vesa's comments. I increase the advertising budget by 20% in a geographic area A and I do not increase it in a geographic area B. Six months later, my sales go up 10% in A and do not go up in B. Was the increase in sales caused by the increase in the advertising budget? The answer yes seems much more reasonable in this setting than in an uncontrolled common sense setting. As we know this is a common way in medical domain to discover the cause or better to say to get evidence for the cause of the event that still needs to be discovered much deeper. The same approach is taken in Machine Learning/ Data Mining with positive and negative training cases where advertising budget was increased, and was not increased along other features (potential multiple other cousants).

Thus the question is: Do we want to try to solve any problem in an uncontrolled common sense setting or we would prefer a more controlled scientific setting?

With best regards

Boris

[bisc-group] Fuzzy Regression: an answer to "What is causality?"

2011-03-18T05:37:00.001-07:00

Dear Lotfi,

I like your raincoat example. In the linear, crisp case, our tool for
approaching this class of questions would be multivariate regression.

The good news is that, in the nonlinear, soft-computing case, there is
an analogous approach. For a good start, the interested reader could
check:

"Fuzzy Regression by Fuzzy Number Neural Networks," James Dunyak and
Donald C. Wunsch II, Fuzzy Sets and Systems, 108 (1), (1999), pp.
49-58.

Related material can also be found in the following two papers:

"A Theory of Independent Fuzzy Probabilities for System Reliability,"
James Dunyak, Donald Wunsch, and Ihab W. Saad, IEEE Trans. On Fuzzy
Systems, v. 7, no. 3, June 1999, pp. 286-294.

"Fuzzy Number Neural Networks," James Dunyak and Donald C. Wunsch II,
Fuzzy Sets and Systems, 112 (3), (2000), pp. 371-380.

I'm grateful for partial support from the National Science Foundation
on all three of these papers, to the co-authors named, and to you for
many stimulating comments and questions that continually make the BISC
discussion group such a fruitful and fascinating venue.

Yours truly,

Don

wunsch@ieee.org

http://people.mst.edu/faculty/dwunsch_profile.html

http://www.linkedin.com/in/wunsch

(573) 341-4521 Office
(573) 341-4532 Fax

Missouri University of Science & Technology
is known as Missouri S&T for short. It was

previously known as the University of Missouri - Rolla (UMR).

From: Lotfi A. Zadeh [mailto:zadeh@eecs.berkeley.edu]
Sent: Monday, March 14, 2011 8:18 PM
To: bisc-group@lists.eecs.berkeley.edu
Subject: [bisc-group] What is causality?

Dear Cathy, Hans, Vesa, Leonid and Stuart:

Thank you for your constructive comments regarding causality. Your
comments shed light on why it is so difficult to formalize a concept
which is so central to human reasoning and cognition.

In my view, existing theories of causality founder on the rocks of
multicausality. In most realistic settings multicausality is the rule
rather than exception. My raincoat's example is typical. For
convenience, let me repeat it. I am a manufacturer of raincoats. I
would like to increase my sales. To this end, I increase the
advertising budget by 20%. Six months later, my sales go up 10%. Was
the increase in sales caused by the increase in the advertising
budget? Can you point to a theory of causality which can answer this
question?

The problem is that the increase in sales may be related to a
multiplicity of causants: rainy weather, better economy, going out of
business of my principal competitor, new managers in my stores, the
increase in advertising, etc., etc., etc. Given the multiplicity of
listed causants, and possibly many others that I cannot think of, it
is not possible to prove or disprove that the increase in sales was
caused by the increase in the advertising budget. Furthermore,
multiplicity of causants necessitates that causality be a matter of
degree.
The question then, is to what degree did the increase in the
advertising cause the increase in sales? In my message of June 30,
2009, I suggested that to answer the question it is necessary to
survey all buyers of my raincoats. Clearly, this is impractical.
Polling of a sample of buyers would provide some fuzzy information but
would not be a proof. No existing theory of causality can address the
raincoat's example. A stronger statement is that it is impossible to
construct a rigorous theory of causality which can deal with
multicausality, except in special cases.

Impossibility of constructing a realistic theory should not come
as a surprise. In a 2009 paper entitled "Toward Extended Fuzzy
Logic--A First Step" (attached), I suggested that there are problems
which do not have provably valid (p-valid) solutions, but may have
fuzzily valid (f-valid) solutions. As an example, I described what I
called the taxi cab problem. I hail a taxi and ask the driver to take
me to address A, in the downtown area. There are two versions. In
version (a), I ask the driver to take me to A the shortest way. In
version (b), I ask the driver to take me to A the fastest way. Version
(a) has a p-valid solution. Version (b) does not have a p-valid
solution because of unpredictability of traffic conditions. In fact,
if the driver had asked me to define what I mean by "fastest way," I
could not do it. The way the driver takes me to A may be viewed as an
f-valid solution based on experience. This simple example suggests
that there are problems for which p-valid solutions do not exist. The
same applies to concepts and theories. In this sense, it is not
possible to formulate a p-valid definition of causality and, ipso
facto, construct a realistic model of causality which can address the
problem of multicausality. In this respect, causality is not unique.
The same can be said about many other concepts, including the concept
of rationality. See John Meech's message of March 27, 2009. Comments
are welcome.

Regards to all.

Lotfi

Lotfi A. Zadeh

Professor in the Graduate School

Director, Berkeley Initiative in Soft Computing (BISC)

Address:

729 Soda Hall #1776

Computer Science Division

Department of Electrical Engineering and Computer Sciences

University of California

Berkeley, CA 94720-1776

zadeh@eecs.berkeley.edu

Tel.(office): (510) 642-4959

Fax (office): (510) 642-1712

Tel.(home): (510) 526-2569

Fax (home): (510) 526-2433

URL: http://www.cs.berkeley.edu/~zadeh/

BISC Homepage URLs

URL: http://zadeh.cs.berkeley.edu/

Top10 books by Janusz Kacprzyk

2011-03-17T04:48:00.001-07:00

From: Madan Gupta <madan.gupta@usask.ca>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Academician Janusz Kacprzyk
Professor, Ph.D., D.Sc.
Fellow of IEEE, IFSA
President of IFSA (International Fuzzy Systems Association)
President of the Polish Society for Operational and Systems Research

Systems Research Institute
Polish Academy of Sciences
ul. Newelska 6
01-447 Warsaw, Poland

Email: kacprzyk@ibspan.waw.pl
http://www.ibspan.waw.pl/~kacprzyk
Google: kacprzyk

Phone: + (48)(22)3810275
Fax: + (48)(22)3810103

Dear Janusz:

Heartiest congratulations for the well deserved recognition for publishing some excellent books in the field of soft computing in general and fuzzy systems in particular!

My good friends like you deserve this type of recognition!!!

Best regards,

Madan

========================================
Dr. Madan M. Gupta
Professor (Emeritus) & Distinguished Research Chair

Director
Intelligent Systems Research Laboratory
College of Engineering
University of Saskatchewan
SASKATOON, SK.
CANADA, S7N 5A9
--------------------------------------------------
Residence:
101 Capilano Court
Saskatoon, SK. S7K 4B9
Canada
--------------------------------------------------
Phone: (Office): 1 - (306) 966 - 5451
(Home) : 1 - (306) 933 - 0663
Fax: (Office): 1 - (306) 966 - 5427
--------------------------------------------------
E-Mail: Madan.Gupta@usask.ca
http://www.usask.ca/~Madan.Gupta
http://www.engr.usask.ca/faculty.php?Madan.Gupta
=================================================

On 2/22/2011 4:57 PM, Etienne Kerre wrote:

Dear Janusz,
My sincere congratulations for this big achievement and for your continuous efforts to promote soft computing in general and fuzzy in particular!
warm regards
etienne

Dear Friends, Dear BISCers,

I wish to share with you my happiness from a great news I have just received from Springer that the book series "Studies in Computational Intelligence", which I am the editor in chief of, is on the second place with respect to the number of chapter downloads, and only LNCS is better. This has been achieved just in a short period of a couple of years.

Many of you have been the authors/editors/contributors to books and volumes in that series and I wish to thank you so much for your great efforts.

I am expecting a further grows of the series with many new books and volumes scheduled, on new and hot topics, by top people, from top univeristies.

Warm wishes and regards

Janusz Kacprzyk

------ Wiadomos'c' oryginalna ------

Temat:     top10

Data:     Mon, 21 Feb 2011 12:15:18 +0100

Nadawca:     Ditzinger, Thomas, Springer DE <Thomas.Ditzinger@springer.com>

Adresat:     Janusz Kacprzyk <Janusz.Kacprzyk@ibspan.waw.pl>

Dear Janusz,

I have just received a statistics about the e-book chapter downloads of

individual book series in 2010:

Among all Springer book series worldwide SCI is on number 2! (see

attachment), only "short" behind LNCS!

Congratulations on this fantastic success!

With the best wishes and smiles

Tom

Thomas Ditzinger

================================================================

Dr. Thomas Ditzinger, Senior Editor, Engineering/Applied Sciences

Springer-Verlag, Tiergartenstr. 17, D-69121 Heidelberg, Germany

Phone: +49-6221-487 8623, Fax: +49-6221-4876 8623,

Skype: tditzinger,http://www.springer.com <http://www.springer.com/>

mailto:Thomas.Ditzinger@springer.com

================================================================

CWW, Scientific Chinese Language -- many methods of locating characters are better than a unique method

2011-03-16T08:32:00.001-07:00

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear all, 18/2/11

At 14:58 -0500 17/02/11, Paul P. Wang wrote:
>
> *********************************************************************
> Berkeley Initiative in Soft Computing (BISC)
> *********************************************************************
>
> Dear Leonid:
>
> I am not sure the accurate answer to your question.I did look at the earliest Oracle writings appeared on the turtle backs as they evolved into ancient Chinese ideographic characters.they do not have neither prefixes,nor the suffixes.
>
> However,one may interpretation the prefixes and suffixes as 'radicals'.Some more complex Chinese words may have several 'radicals'.The problem associated with such interpretation is that the 'radicals' physically may appear in the following possible combinations;right-left,top-bottom,or the combinations of 'right ,left ,top ,bottom.',as the result of very heavy price to pay being 'Ideographs'.
>
> You also have touched upon a very sensitive nerve of the weakness of Chinese language.

Once again, one should now confuse languages with writing systems.

> Through out the history,many researchers have attempted to come up with a unique way to locate a word in the Chinese dictionary.In my opinion,none of them has been successful! Contrary to other language like English,this has never been a problem.
>
> Hence, the intended effort of Scientific Chinese Language reform MUST set a goal of achieving this very basic requirement!

Why a UNIQUE access system? This "MUST", in capitals, seems ill-founded.

It is far better to propose to users several systems, according to
their various competences.

There are many systems to locate Chinese or Japanese ideograms in
dictionaries. To lookup characters by radicals is a common way but not
the only one. The majority use the traditional 214 radicals, but
Hadamitsky uses a reduced set of 72 radicals in his dictionary of
Japanese compounds.

Note that, from a study done in 1989 by F. Tcheou on the Ci Hai (see
of words) dictionary of 14872 characters (the total number since the
origins is about 67000), see
http://www.aclclp.org.tw/rocling/1990/K05.pdf, almost all Chinese
characters can be decomposed as the graphical superposition of one or
more radicals (one can however suspend the writing of a radical to
write another).

There are also form-based methods which avoid the pitfall that many
characters have more than 1 candidate radical (for search purposes) in
them, or that the historical radical is not visible in the
contemporary form of the character. For example, there is a very
efficient such method in the "Kanji finder index". Another one has
been introduced by Jack Halpern and is proposed (alongside the
radical-based and pronunciation-based access methods) in the Kodansha
Kanji Learner's Dictionary (the first version was published in 1990 as
"The New Japanese-English Character Dictionary"). It has about 3500
characters and it takes typically a few seconds only to locate a
character.

And, at least for Japanese, access by the pronunciation(s) -- at least
by the less ambiguous, often polysyllabic "kun yomi" (coming from the
original Japanese, as opposed to monosyllabic "on yomi", coming from
Chinese) --, is a useful option (see for example Vaccari's
dictionaries).

With best regards,

Ch.Boitet

> Respectfully yours
>
> Paul
>
> On Mon, 14 Feb 2011, Perlovsky, Leonid Civ USAF AFMC AFRL/RYHE wrote:
>
>> *********************************************************************
>> Berkeley Initiative in Soft Computing (BISC)
>> *********************************************************************
>>
>
> --
> Professor Paul P.Wang
> Department of Electrical & Computer Engineering
> Pratt School of Engineering
> Duke University,Durham ,NC,27708 ,USA
>
> ppw@ee.duke.edu
> voice:919 660 5259
> fax:919 660 5293

GETALP, LIG-campus, BP 53 (ex: GETA, CLIPS, IMAG-campus)
Tel: +33 (0)4 76 51 43 55 / 51 48 17 Fax: +33 (0)4 76 63 56 86
385, rue de la Bibliothe`que Mel: Christian.Boitet@imag.fr
38041 Grenoble Cedex 9, France

Congratulations/ Kacprzyk

2011-03-16T08:31:00.001-07:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>

Dear Janusz:

Congratulations on the success of your Series. You have an
uncanny ability to make
right decisions.

With warmest regards.

Sincerely,

Lotfi

--
Lotfi A. Zadeh
Professor in the Graduate School
Director, Berkeley Initiative in Soft Computing (BISC)

BISC Homepage URLs
URL: http://zadeh.cs.berkeley.edu/

[bisc-group] http://japansupport.blogspot.com/ We have received many replies for our recent email about our warm condolences to the victims of the catastrophic earthquake and ensuing tsunami in Japan

2011-03-16T08:29:00.001-07:00

Dear BISC

We have received many replies for our recent email about our "warm
condolences to the victims of the catastrophic earthquake and ensuing
tsunami in Japan"

It is not possible to upload them in our web site as we did in the
past in similar cases.
So, we have created a new forum http://japansupport.blogspot.com/ for
all of you
from where everybody can donate to American Red Cross,

Do you want to be in our scientific committee?
http://japansupport.blogspot.com/

If yes, because, we do not wish to add names in the committee without
their permission
please, put a link from your web site(s) to
http://japansupport.blogspot.com/
and then inform us by email
Many Thanks
Rena Politi

Condolence to our Japanese fellow

2011-03-16T08:28:00.001-07:00

At this hour of sudden death of many Japanese people in the tragedy of
earthquake and aftermath tsunami, I want to share my heartfelt condolences
with all of you and specially with our Japanese fellow in BISC.

I lived and studied in Sendai, near the epicenter of the earthquake, for 3
years. Sendai is well-known in Japan as The Green City, due to its
extraordinary beautiful environment and scenery, both in coastal and
mountain area together with the city itself. It is very sad to see how these
beautiful area and those lovely Japanese people suffer such enormous pain.
They've been through the roughest moments anybody could imagine. Our
thoughts and prayers are with them. May God bless soul of those passed away
in this tragedy.

Sincerely
Alireza Fatehi

------
Assistant Professor
Advanced Process Automation & Control (APAC) research group
Faculty of Electrical Eng.
K. N. Toosi University of Technology
Tel:+98-21-84062207 // Fax: +98-21-88462066
email: fatehi@kntu.ac.ir
http://saba.kntu.ac.ir/eecd/fatehi

http://japansupport.blogspot.com

What is causality?

2011-03-16T08:27:00.001-07:00

Dear Cathy, Hans, Vesa, Leonid and Stuart:

Thank you for your constructive comments regarding causality. Your
comments shed light on why it is so difficult to formalize a concept
which is so central to human reasoning and cognition.

Regards to all.

Lotfi

--
Lotfi A. Zadeh
Professor in the Graduate School
Director, Berkeley Initiative in Soft Computing (BISC)

BISC Homepage URLs
URL: http://zadeh.cs.berkeley.edu/

Fwd: [bisc-group] don't mix languages and writing systems: CWW and Languages

2011-02-04T00:51:00.001-08:00

At 23:29 -0700 26/01/11, yingxu@ucalgary.ca wrote:

Content-Type: text/plain; charset="UTF-8"; name="message-header.txt"
Content-Disposition: inline; filename="message-header.txt"
Content-Transfer-Encoding: 8bit
MIME-Version: 1.0
X-Mailer: MIME-tools 5.427 (Entity 5.427)

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Paul, Yu-Chi, and all,

With regard to comparative studies between the two basic categories of
languages, i.e., alphabetic and ideographic languages, I've some
observations in cognitive informatics as follows. By following Lotfi's
convention, I'd denote the former as Type 1 languages and the latter Type
2 languages.

one should not confuse languages with writing systems.

One cannot characterize a language by its writing system, because many languages have several writing systems, and some none.

For exemple, Mongolian has a Mongolian, a Cyrillic, and a Chinese writing system. Some languages of central Asia have 4. Vietnamese was written for centuries in the Chinese writing system, although it is in the family of Thai and Lao. Now it is written in an alphabetic writing system based on the Roman script, not on Pali-related scripts like Thai and Lao.

Some languages have no writing system, e.g. Japanese in the 2nd century, or (still) some of the 6000 or so languages still around. Hence they could not be of type I or II, right?

Languages are first characterized as spoken, not written means of communication.

Maybe the whole discussion should be rephrased in terms of writing systems rather than languages, or combinations of <WS, LG>.

Best regards,

Ch.Boitet

a) The mathematical models of Type 1 languages are sequential
combinations; while those of Type 2 languages are structural combinations.

b) The most significant physiological contribution of early-phase language
acquisition to the brain is the development of memory, which forms the
foundation, as well constraints, of the natural intelligence and human
inference abilities. Although all languages are equivalent in function,
their impacts to the development of the brain would be different. Type 1
languages require more and help to advance short-term memory (STM) because
the combinational complexity is higher in sentence generation; while Type
2 languages need more long-term memory (LTM), therefore less STM, because
words are wired as firmware unit in LTM. These are the different effects
of language development that influence the configurations of the brains of
these two people who write/speak different types of languages.

c) As a consequence, Type 2 language users are more efficient in language
manipulation once they are acquired, and also much shorter in writing and
speaking; While Type 1 language users are less efficient in sentence
generation and need a longer thread in reasoning on the same ideal as that
in Type 2. Therefore, in term of thinking speed influenced by languages,
Type 2 is faster than that of Type 1 in general.

d) However, intelligence is not merely dependent on reasoning speed. The
formation of a larger STM, as a result of Type 1 language acquisitions in
the early development of the brain, enables Type 1 language users to
posses a wider and dynamic inference space in STM for complex problem
solving and creative activities. While the advantage of Type 2 language
users, with a larger LTM, is helpful for cumulative and stable knowledge
acquisition and systematical problem solving.

e) Bilinguals and multilinguals, particularly those with a combination of
Type 1 and Type 2 languages, may take advantage in both STM and LTM
development.

f) The above phenomena and reasoning are based on the impacts of early
language development to the brain. Though, late language development of
adults in acquisition of professional languages and abstract notations,
such as mathematics and symbolic systems, has equivalent effect to brain
development as those of natural languages.

Happy Chinese New Year to all!

Yingxu
---------------------------------------------------------------------------
Yingxu Wang, PhD, P.Eng, F.WIF, SMIEEE, SMACM
Professor of Cognitive Computing and Software Engineering
Visiting Professor: Oxford Univ. (1995), Stanford Univ. (2008),
UC Berkeley (2008)
Director, International Institute of Cognitive Informatics and
Cognitive Computing (IICICC)
Director, Theoretical and Empirical Software Engineering
Research Center (TESERC)
Dept. of Electrical and Computer Engineering
Schulich School of Engineering
University of Calgary
2500 University Drive, NW
Calgary, AB, Canada T2N 1N4
Tel: (403) 220 6141
Fax: (403) 282 6855
http://enel.ucalgary.ca/People/wangyx
http://enel.ucalgary.ca/People/wangyx/Research/Research.htm
http://enel.ucalgary.ca/IJCINI
http://www.igi-global.com/journals/details.asp?id=7981 (IJSSCI)
http://www.ucalgary.ca/icci_cc2011/
---------------------------------------------------------------------------
Editor-in-Chief, International Journal of Cognitive Informatics and
Natural Intelligence (IJCINI)
Editor-in-Chief, International Journal of Software Science and
Computational Intelligence (IJSSCI)
Associate Editor, IEEE Transactions on System, Man, and
Cybernetics (Part A)
Associate Editor, Journal of Advanced Mathematics and
Applications (JAMA)
Associate Editor in Chief, International Journal of Applied Metaheuristic
Computing (IJAMC)
Chair, Steering Committee of IEEE ICCI Conference Series
Editor-in-Chief, CRC Book Series in Software Engineering
---------------------------------------------------------------------------

---------- Forwarded message ----------
Date: Sat, 22 Jan 2011 13:00:29 -0500 (EST)
From: Paul P. Wang <ppw@ee.duke.edu>
Cc: ppw@ee.duke.edu
Subject: CWW,Scientific Chinese Language----Talkinmg Point No.23,
Chinese Alphabets

Alphabets in a nature language represent its phonetics and the Chinese
language is generally regarded as not an alphabetic language.The English
language has 26 alphabets [same as Classical Roman alphabets except the
Classical Roman language does not have J,U and W and the Classical Roman
alphabets are in upper case only].

... [Sorry, there is a limit to message size.]

Since the language is a system of relations of parts constituting the
signs that organize the action necessary for the self-maintenance of a
living system as stressed by J.Z.Young,I feel we can not ignore its
importance and perhaps to extend the CWW,Computing With Words,also to the
basic level of the words themselves.

--
Professor Paul P.Wang
Department of Electrical & Computer Engineering Pratt School of
Engineering Duke University,Durham ,NC,27708 ,USA

ppw@ee.duke.edu
voice:919 660 5259

> fax:919 660 5293

--
-------------------------------------------------------------------------
Christian Boitet
(Pr. Universite' Joseph Fourier)
Groupe d'Etude pour la Traduction Automatique
et le Traitement Automatisé des Langues et de la Parole
G E T A L P

GETALP, LIG-campus, BP 53 (ex: GETA, CLIPS, IMAG-campus) Tel: +33 (0)4 76 51 43 55 / 51 48 17 Fax: +33 (0)4 76 63 56 86
385, rue de la Bibliothe`que Mel: Christian.Boitet@imag.fr 38041 Grenoble Cedex 9, France

Fw: Dear Professor Lien, Your papers will appear in ISI, EI Compendex, IEEEAM, SCOPUS, ACM, WSEAS E-Library, IET (IEE), ASM, ACS, CSA, ELSEVIER , ZENTRALBLATT, MATHSCINET, DPP,EI, CSBA, Ulrigh, DEST, EBSCO, EMBASE, GEOBASE, BIOBAS

2011-01-19T02:07:00.001-08:00

----- Original Message -----

From: Pei-te Lien

To: Maria Makrynaki

Sent: Tuesday, January 18, 2011 8:30 PM

Subject: Re: Dear Professor Lien, Your papers will appear in ISI, EI Compendex, IEEEAM, SCOPUS, ACM, WSEAS E-Library, IET (IEE), ASM, ACS, CSA, ELSEVIER , ZENTRALBLATT, MATHSCINET, DPP,EI, CSBA, Ulrigh, DEST, EBSCO, EMBASE, GEOBASE, BIOBAS

Sorry. I possess no expertise nor interest in your conferences. Thanks for removing me from your mailing list. Pei-te Lien

On Tue, Jan 18, 2011 at 3:51 AM, Maria Makrynaki <wseas@mail.gr> wrote:

Dear Professor Lien,

We had invited you (in Dec.16, 2010) as INVITED SPEAKER
in the following conferences : (see the particular conference below)
Will you come? Will you upload your paper? Do you need 20 additional days to prepare your paper?
Your paper will be published in ISI Books, Proceedings and Journals
and you will receive this material in the conference

CONFERENCES:

Lanzarote, Canary Islands, Spain, May 27-29, 2011

International Conference on Energy, Environment, Entrepreneurship, Innovation (ICEEEI'11)

International Conference on Social Science, Social Economy and Digital Convergence (IC-SSSE-DC'11)

International Conference on Manufacturing, Commerce, Tourism and Services (ICM-CTS'11)

Faculty of Law, Transilvania University of Brasov, Romania, April 7-9, 2011

International Conference on TRADITION and INNOVATION in the ACADEMIC LAW SYSTEM (TIALS '11)

International Conference on TECHNOLOGY POLICY and LAW (TPL '11)

International Conference on PRIVATE LAW (PL '11)

International Conference on LIABILITY and SCIENTIFIC ENQUIRY (LSE '11)

International Conference on INTELLECTUAL PROPERTY AND INFORMATION MANAGEMENT (IPM '11)
Delegates participating in these conference will be invited to upload their papers in many international journals , Springer Verlag journals, Hindawii, NOVA, Publishers, NAUN, University Press and IEEE.AM(some of them in ISI)

We have ensured that all the WSEAS Publications are indexed and included in
ISI, EI Compendex, IEEEAM, SCOPUS, ACM, WSEAS E-Library, IET (IEE), ASM, ACS, CSA, ELSEVIER , ZENTRALBLATT, MATHSCINET, DPP,EI, CSBA, Ulrigh, DEST, EBSCO, EMBASE, GEOBASE, BIOBASE, BIOTECHNOBASE.

* INVITED PAPER means that your paper will be published in our ISI books and Journals

You must upload your paper via the conference site
See them via www.wseas.org

Select the conference that you like and then upload your paper
via the Link: Submit a Paper

The Invited Speakers has the additional privilege of additional Journal Publication after the conference

* As in 2010, (see here 2010), all the WSEAS Publications in Chemistry and Chemical Engineering will be included in ISI, EI Compendex, IEEEAM, SCOPUS, ACM, WSEAS E-Library, IET (IEE), ASM, ACS, CSA, ELSEVIER , ZENTRALBLATT, MATHSCINET, DPP,EI, CSBA, Ulrigh, DEST, EBSCO, EMBASE, GEOBASE, BIOBASE,BIOTECHNOBASE.

* As in 2010, (see here 2010), all the WSEAS Publications in Chemistry and Chemical Engineering will be archived in ISI, EI Compendex, IEEEAM, SCOPUS, ACM, WSEAS E-Library, IET (IEE), ASM, ACS, CSA, ELSEVIER , ZENTRALBLATT, MATHSCINET, DPP,EI, CSBA, Ulrigh, DEST, EBSCO, EMBASE, GEOBASE, BIOBASE, BIOTECHNOBASE.

* As in 2010, (see here 2010), al the Authors will be invited to submit extended versions to a NAUN Journal or University Press

* As in 2010, (see here 2010), in collaboration with the Institute of EUROPMENT, IEEEAM, IASME, IARAS, WSEAS.

The ISI update its data base
http://science.thomsonreuters. com/m/excel/conference_titles_ 1998-2010.xlsx
every July and every January
The last time that updated it was July 2010,
As I was informed they appended in their data base
all the WSEAS Conferences until February 2010.
They will update it again after Febr. 2011
So, if your conference was after February 2010, check again the data base of ISI in January of 2011
http://science.thomsonreuters. com/m/excel/conference_titles_ 1998-2010.xlsx

Yours Sincerely

Maria Makrynaki

The WSEAS

If you want to un~subscribe, send an email to wseas-team@wseas.org with the following command as
Subject: un~subscribe plien@polsci.ucsb.edu

If you have more than one emails,
send an email to wseas-team@wseas.org with the following command as
Subject: un~subscribe plien@polsci.ucsb.edu, otheremail1, otheremail2,...

ANSWER Fwd: [bisc-group] Probability vs. Fuzziness/ Paul Snow

2011-01-15T01:03:00.001-08:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>
Date: 15 January 2011 03:40
Subject: [bisc-group] Probability vs. Fuzziness/ Paul Snow
To: Paul Snow <paulusnix@gmail.com>

Dear Paul,

    Many thanks for your as always thoughtful response to my message. In your message, you assert that the meaning of the proposition, p: Most Swedes are tall, may be represented through the use of standard probability theory. Could you describe how it can be done? You also assert that the probability that Magnus is short can be handled through the use of standard probability theory. Again, how would you do it?

    The principal point of my earlier message was that standard probability theory cannot deal with problems stated in a natural language. You suggest that my problems are ill-posed. This is conventional wisdom. In reality, the problems are well-posed if you have at your disposal the machinery of Computing with Words (CW or CWW). There are many misconceptions about what CW is and what it has to offer. Following is a clarification.

    Computing with Words is a system of computation in which the objects of computation are words, phrases and from a natural language. The carriers of information are propositions. It is important to note that Computing with Words is not related to natural language processing. It is also important to note that Computing with Words is the only system of computation which offers a capability to compute with information described in a natural language. A prerequisite for computation is precisiation of meaning. In Computing with Words, the meaning of a proposition is precisiated by expressing meaning as a restriction (constraint) on the values which a variable is allowed to take.

    The point of departure in CW is what is called an information set, I. I is a collection of propositions at least one of which is drawn from a natural language. The terminus is an answer to a question q, of the form: What is the value of a variable Y? The information set I, serves as a basis for computation of the answer, Ans(q/I). Generally, Ans(q/I) is not a value of Y but a restriction (constraint) on the values which Y is allowed to take. Equivalently, Ans(q/I) identifies those values of Y which are consistent with I. In CW, consistency is equated to possibility, with the understanding that possibility is a matter of degree.

    Computation of Ans(q/I) involves two phases. In Phase 1, propositions in I are precisiated through the use of so-called generalized-constraint-based semantics, GCS. In GCS, the meaning of a proposition drawn from a natural language is represented as a generalized constraint. The principal generalized constraints are possibilistic, probabilistic and veristic. In the realm of natural languages, constraints are preponderantly possibilistic. In Phase 2, the generalized constraints which represent the meaning of precisiated propositions, serve as objects of computation, with computation involving propagation of generalized constraints. The principal rule governing propagation of constraints is the extension principle of fuzzy logic. Basically, the extensions principle is a formalism which propagates restrictions on the values of arguments of a function, f, to a restriction on the value of f.

    What is important to note is that CW is the only system of computation which offers a capability to compute with information described in a natural language. That is why the problems posed in my message cannot be solved through the use of standard probability theory. However, given the inertia and deep entrenchment of standard probability theory, this is a point that may take a while to gain wide acceptance.

    With my warm regards.

    Sincerely,

    Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

Fwd: [bisc-group] Re: Probability vs fuzziness/ Snow, Norvig

2011-01-15T01:00:00.001-08:00

From: Paul Snow <paulusnix@gmail.com>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Hello, again, Lotfi:

Thank you for an invigorating exchange.

> Note that the machinery that is used is far removed from standard probability theory. Kolmogorov's axioms are of little use.

Not so. You have relied heavily on standard probability theory for a
crucial step in your analysis.

I don't object to your removal of the most urgent probabilisitic
concern, how Magnus was selected, by baldly assuming a uniform
sampling distribution over some population to which he belongs.

That is, however, a use of standard probability theory, one
axiomatization of which is Kolmogorov's. Except for that use, all you
would have is a population average, with no definite relationship to
Magnus' individual situation at all.

I also don't see what "machinery" has to do with it. Possibilists
don't use the machinery of probability theory, but we now know that
they routinely restate ordinary calculations within the scope of
standard probability theory. Any other example of one formalism
offering a semantics for a different formalism would reveal the danger
of relying on "machinery" for claiming incompatibility among
formalisms.

And, of course, the "machinery" by which even a thoroughly standard
probabilist arrives at numerical probability estimates in any
particular application domain might have little to do with probability
theory, but instead much to do with the domain theory in question.

> Standard probability theory cannot get beyond this point because there are no means of precisiating the meaning of "Most Swedes are tall."

On the contrary, "standard probability theory," along with any number
of other things, gets beyond this point because there are so many
means of precisiating some plausible meaning of "Most Swedes are
tall."

So, that takes us back to the point where I dropped into this conversation.

You propose to answer to the question "What is the probability that
Magnus is short?" consistent with some other statements. Is yours a
responsive answer to the question asked?

The query, as it is stated, is ill-posed. You make assumptions to
rehabilitate the query so that it is well-posed, with respect to one
particular formalism. Somebody else might make other assumptions, and
successfully rehabilitate the query with respect to a different
formalism.

Which formalism, if either, has actually provided the information
sought by the person who made the query in the first place? Well, if
you're the person who made the query, then that simplifies matters :)
. But if somebody else is asking, then things aren't so clear.

I have long been persuaded that yours is a fine answer. But since
yours isn't the only possible principled answer, I think comparative
semantic investigation and explorations of different axiomatic
foundations are warranted expenditures of effort.

Best wishes,

Paul

Fwd: [bisc-group] Probability vs fuzziness/ Snow, Norvig

2011-01-12T00:35:00.001-08:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>
Date: 12 January 2011 04:28
Subject: [bisc-group] Probability vs fuzziness/ Snow, Norvig
To: bisc-group@lists.eecs.berkeley.edu
Cc: Paul Snow <paulusnix@gmail.com>, Peter Norvig <pnorvig@google.com>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Paul:

    Many thanks for your thoughtful response to my message. Your response makes it clear where we agree and where we don't. The problem which you chose to discuss is a good problem for this purpose.

    To see the problems which I posed in a proper perspective, what should be recognized is that they fall within the province of Computing with Words (CW or CWW). In CW, the point of departure is what is called an information set, I . I consists of a collection of propositions at least one of which is drawn from a natural language.The terminus is an answer to a question, q: What is the value of a variable Y? In general, the answer, Ans(q/I), is not a value of Y but a restriction(constraint) on the values which Y can take. Equivalently, Ans(q/I) may be interpreted as the values of Y which are consistent with I. To solve the problems which I posed, what is needed is the machinery of CW. Without this machinery the problems make no sense. If they were included in a final examination in a course on probability theory, no student would be able to solve them.

    Turning to the problem under discussion, I deliberately did not indicate how Magnus is selected. As you point out, without this information the probability that Magnus is short is between 0 and 1.

    Let us assume that Magnus is chosen at random. What, then, is the probability that he is short, given that most Swedes are tall? Standard probability theory cannot get beyond this point because there are no means of precisiating the meaning of "Most Swedes are tall." Note that the answer to the question is not expected to be a number such as 0.13. What is expected is an answer of the form "low," with the understanding that "low" is a fuzzy set whose membership function is the terminus of computation. In this sense, there is a unique correct answer, given I.

    What is the meaning of "Most Swedes are tall?" To answer this question what are needed are two basic concepts in fuzzy logic: (a) the concept of cardinality (count) of a fuzzy set; and (b) the concept of the probability measure of a fuzzy event. Neither of these concepts lies within the conceptual structure of standard probability theory.

    With regard to the concept of cardinality, assume for simplicity that A is a fuzzy set in a finite space, U = {u₁,...u_n} with µ_A(u_i), i=(1,...n), representing the grade of membership of u_i in A. The cardinality (count) is defined as the sum of the µ_i. The probability of a fuzzy event, X is A, is defined as the sum µ₁p₁+...+µ_np_n, where p_i is the probability of u_i.

    The meaning of "Most Swedes are tall," may be represented as Proportion(tall.Swedes/Swedes) is most, where the fuzzy quantifier, most, is represented as a fuzzy number with membership function µ_most. Let P=[Name;Height] be a population of n Swedes, with the height of name_i being h_i, i=1,....n. The proportion, r, of tall Swedes among Swedes may be expressed as r=1/n(µ_tall(h₁)+...+µ_tall(h_n)), where µ_tall is the membership function of tall. Thus, the precisiated meaning of "Most Swedes are tall" is the fuzzy constraint: r is most. Similarly, the proportion of short Swedes among Swedes may be expressed as s=1/n(µ_short(h₁)+...+µ_short(h_n)), where µ_short is the membership function of short. This is an instance of a key idea in CW, namely, that the precisiated meaning of a proposition drawn from a natural language may be represented as a constraint.

    At this point, the posed problem reduces to the following variational problem: Given that r is most, what is the restriction on the values of s? The answer to this question is provided by the extension principle. More specifically,

    µ_s(v)=sup_h1,...,hnµ_most(1/n(µ_tall(h₁)+...+µ_tall(h_n))

subject to

    v=1/n(µ_short(h₁)+...+µ_short(h_n))

Since Magnus is picked at random, the probability that Magnus is short is s, the proportion of short Swedes among Swedes, with the understanding that s is a fuzzy set.

    In summary, what is described above is a CW-based solution to the problem. Note that the machinery that is used is far removed from standard probability theory. Kolmogorov's axioms are of little use.

Dear Peter,

    In your message, you asked for fuzzy logic-based solutions to the problems which I posed. The above should give you an idea. Other problems may be solved similarly. The answer to Problem 7 is: h_ave=most×tall+(1-most)×h_min.

    Warm regards,

    Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

Probability vs fuzziness/ Snow

2011-01-11T00:07:00.001-08:00

From: Paul Snow <paulusnix@gmail.com>

Dear Lotfi,

Thank you for the kind words. I hope you've had great holidays.

You wrote, "What I discern in what you wrote is that you remain
unconvinced that probability theory is in need of generalization."

No. I would be disappointed if probability did not continue to
progress. Multivalued logics suggest many directions for fruitful
future exploration.

Where you and I have disagreed in the past about examples like

"Most Swedes are tall. Magnus is a Swede. What is the probability that
Magnus is short?"

is that you have felt it is a shortcoming of "probability theory" not
to furnish some unique "correct answer" to such a query.

In contrast, I think that that is a strength. If we look at the
problem from any of the "objective" probability viewpoints, not only
are we told, correctly, that the information is insufficient for
inference, but we can see what we lack. Quite apart from any fuzziness
about the labels that appear in the problem, there is the urgent
problem of how Magnus has been selected from among the Swedes for our
scrutiny.

Not bad, then, for a formalism not only to avoid the shallowness of
furnishing some "answer" regardless of whatever input it is fed, but
also to provide guidance on what more it needs to produce something
interesting and reliable.

But, of course, another reason why "probability theory" doesn't
provide a unique answer is that there is more than one interpretation
of probability itself. Perhaps we are operating within one of the
"subjective" interpretations, and we are modeling someone's beliefs.

The standard of correctness, in that case, is solely whether the
probability model faithfully describes that person's beliefs. The
range of human beliefs that are consistent with the information
provided in the example are legion. Of special relevance to this
problem, one person might have an opinion about how Magnus was chosen,
the next person not.

So, we have a riotous profusion of potential probabilistic answers, in
addition to the "insufficient information" heard from someone whose
beliefs resemble the objectivist perspective on the query. Of those
who venture an opinion, anything from zero through one might he heard,
depending on beliefs about the unspecified sampling procedure - even
without considering the many approaches probabilists might adopt for
the labeling issues.

None of which, so far can as I can see, entails any lack of enthusiasm
for your project of generalizing probability theory to operate from
fuzzy foundations. If anything, my appreciation of how far the theory
has come from its original obsession with drawings from an urn can
only heighten my anticipation of even more glorious progress to come.

Best wishes and warm thoughts,

Paul

Fwd: [bisc-group] Probability vs fuzziness/ Snow

2011-01-10T23:54:00.001-08:00

From: Farley Nobre <fsmnobre@gmail.com>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Lotfi,

this true specially for the decision problems whereas:

(a) Alternatives of choice are not simply given but they must be generated through a process of search.

(b) The probability distributions of outcomes are unknown and may be only estimated through high computational costs.

(c) Humans manipulate natural concepts (Bernstein et al. 1997). Therefore, most of the uncertainty that pervades the alternatives and their consequences are classified into vagueness (Black 1937, 1963) and fuzziness (Zadeh 1965, 1973) rather than probabilistic uncertainty (IEEE 1994).

(d) Satisfactory outcomes are preferable to maximization or optimization (March & Simon, 1958).

REF.: http://www.springerlink.com/content/3141862146981uh3/ (pp. 373)

Best wishes, Farley

Farley Simon Nobre

Professor of Knowledge Management and Innovation

School of Management / Federal University of Parana - Curitiba, Brazil

http://www.innovationtech.com.br/nobre.html

2011/1/4 Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Paul:

    Thank you for your as always erudite comment. What I discern in what you wrote is that you remain unconvinced that probability theory is in need of generalization. In my view, it is. More specifically, I feel that the problem-solving capability of probability theory can be substantially enhanced through addition to its armamentarium of concepts and techniques drawn from fuzzy logic. My position is articulated in my 2002 paper entitled, "Toward a perception-based theory of probabilistic reasoning with imprecise probabilities," published in a special issue on imprecise probabilities in the Journal of Statistical Planning and Inference. In this paper, I suggested a generalization of probability theory. In commenting on the papers in the special issue, this is what the co-editors had to say:

"There is a wide range of views concerning the sources and significance of imprecision. This ranges from de Finetti's view, that imprecision arises merely from incomplete elicitation of subjective probabilities, to Zadeh's view, that most of the information relevant to probabilistic analysis is intrinsically imprecise, and that there is imprecision and fuzziness not only in probabilities, but also in events, relations and properties such as independence. The research program outlined by Zadeh is a more radical departure from standard probability theory than the other approaches in this volume." (Jean-Marc Bernard)

"I think that your ideas on perception-based probability are exciting and I hope that they will be published in probability and statistics journals where they will be widely read. I think that there is an urgent need for a new, more innovative and more eclectic, journal in the area. The established journals are just not receptive to new ideas - their editors are convinced that all the fundamental ideas of probability were established by Kolmogorov and Bayes, and that it only remains to develop them!" (Peter Walley)

    Since you are content with standard probability theory, how would you deal with the following simple questions and problems.

Unemployment in California stands at 12.6%. Robert lives in California. What is the probability that Robert is unemployed? What is the possibility that Robert is unemployed?

Robert is tall. What is the probability that Robert is short?

It is very likely that Robert is tall. What is the probability that Robert is short?

X is much larger than 5. What is the probability that X is larger than 50?

Most Swedes are tall. Magnus is a Swede. What is the probability that Magnus is short?

Most Italians are not very tall. Magnus is much taller than most Italians. How tall is Magnus?

Most Swedes are tall. What is the average height of Swedes?

X is a real-valued random variable. Usually, X is much larger than approximately a. Usually X is much smaller than approximately b. What is the probability that X is approximately c?

    I am looking forward to your solutions of some or all of these problems. Members of the BISC Group are welcome to comment.

    With warm regards.

    Sincerely,

    Lotfi

Fwd: [bisc-group] Probability vs fuzziness/ Snow

2011-01-05T02:41:00.001-08:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>

Dear Paul:

Thank you for your as always erudite comment. What I discern in what you wrote is that you remain unconvinced that probability theory is in need of generalization. In my view, it is. More specifically, I feel that the problem-solving capability of probability theory can be substantially enhanced through addition to its armamentarium of concepts and techniques drawn from fuzzy logic. My position is articulated in my 2002 paper entitled, "Toward a perception-based theory of probabilistic reasoning with imprecise probabilities," published in a special issue on imprecise probabilities in the Journal of Statistical Planning and Inference. In this paper, I suggested a generalization of probability theory. In commenting on the papers in the special issue, this is what the co-editors had to say:

"There is a wide range of views concerning the sources and significance of imprecision. This ranges from de Finetti's view, that imprecision arises merely from incomplete elicitation of subjective probabilities, to Zadeh's view, that most of the information relevant to probabilistic analysis is intrinsically imprecise, and that there is imprecision and fuzziness not only in probabilities, but also in events, relations and properties such as independence. The research program outlined by Zadeh is a more radical departure from standard probability theory than the other approaches in this volume." (Jean-Marc Bernard)

"I think that your ideas on perception-based probability are exciting and I hope that they will be published in probability and statistics journals where they will be widely read. I think that there is an urgent need for a new, more innovative and more eclectic, journal in the area. The established journals are just not receptive to new ideas - their editors are convinced that all the fundamental ideas of probability were established by Kolmogorov and Bayes, and that it only remains to develop them!" (Peter Walley)

Since you are content with standard probability theory, how would you deal with the following simple questions and problems.

Unemployment in California stands at 12.6%. Robert lives in California. What is the probability that Robert is unemployed? What is the possibility that Robert is unemployed?
Robert is tall. What is the probability that Robert is short?
It is very likely that Robert is tall. What is the probability that Robert is short?
X is much larger than 5. What is the probability that X is larger than 50?
Most Swedes are tall. Magnus is a Swede. What is the probability that Magnus is short?
Most Italians are not very tall. Magnus is much taller than most Italians. How tall is Magnus?
Most Swedes are tall. What is the average height of Swedes?
X is a real-valued random variable. Usually, X is much larger than approximately a. Usually X is much smaller than approximately b. What is the probability that X is approximately c?

    I am looking forward to your solutions of some or all of these problems. Members of the BISC Group are welcome to comment.

    With warm regards.

    Sincerely,

    Lotfi

[bisc-group] Statistics of citations

2011-01-04T11:47:00.001-08:00

---------- Forwarded message ----------
From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>
Date: 4 January 2011 02:28
Subject: [bisc-group] Statistics of citations
To: bisc-group@lists.eecs.berkeley.edu

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Members of the BISC Group:

    Dr. Ronald P. Loui is an expert in statistics of citations. At my request, he compiled information regarding the twenty highest cited papers--based on Google Scholar--in Information and Control, Information Sciences and IEEE Transactions on SMC. For your information, following is a part of his findings. Ronald deserves a round of applause.

    Regards to all.

    Lotfi

SCHOLAR.GOOGLE.COM Dec 2010 (years are not always accurate)  INFORMATION AND CONTROL

26020 Fuzzy sets, LA Zadeh, 1965
2604 Language identification in the limit, EM Gold, 1967
1601 The rate-distortion function for source coding with side
information at the decoder 3-II: General sources, AD Wyner, 1978
1064 A formal theory of inductive inference. Part I,+, RJ Solomonoff, 1964
905 Process algebra for synchronous communication, JA Bergstra, 1984
869 On certain formal properties of grammars, N Chomsky, 1959
806 A definition of a nonprobabilistic entropy in the setting of fuzzy
sets theory, A De Luca, 1972
759 A new approach to clustering, EH Ruspini, 1969
685 Inductive inference of formal languages from positive data, D Angluin, 1980
672 The definition of random sequences, P Martin-Lof, 1966
616 Relational queries computable in polynomial time, N Immerman, 1986
610 Resolution of composite fuzzy relation equations, E Sanchez, 1976
585 Recognition and parsing of context-free languages in time n3, DH
Younger, 1967
501 Complexity of automaton identification from given data, EM Gold, 1978
476 Probabilistic automata, MO Rabin, 1963
448 A procedure for ordering fuzzy subsets of the unit interval, RR Yager, 1981
443 On a class of error correcting binary group codes, RC Bose, 1960

IEEE SMC

7227 Fuzzy identification of systems and its applications to modeling
and control, T Takagi and M Sugeno, 1985
4648 Outline of a new approach to the analysis of complex systems and
decision processes, LA Zadeh, 1973
3431 Fuzzy logic in control systems: fuzzy logic controller, CC Lee, 1990
2376 On ordered weighted averaging aggregation operators in
multicriteria decisionmaking, RR Yager, transactions on 1988
1690 Optimization of control parameters for genetic algorithms, JJ
Grefenstette, on 1986
1258 State-space and multivariable theory, HH Rosenbrock, 2007
1206 System dynamics: a unified approach, D Karnopp, R Rosenberg, 2007
1155 Scene labeling by relaxation operations, A Rosenfeld, RA Hummel, 2007
1143 Perturbation methods in applied mathematics, JD Cole, 2007
1110 Adaptive probabilities of crossover and mutation in genetic
algorithms, M Srinivas, 2002
1037 Bidirectional associative memories, B Kosko, 2002
953 Textural features corresponding to visual perception, H Tamura, S Mori, 2007
927 Introduction to the Theory of Fuzzy Subsets-vol. 1: Fundamental
Theoretical Elements, A Kaufmann, 2007
846 Shape discrimination using Fourier descriptors, E Persoon, 2007

INFORMATION SCIENCES

5248 The concept of a linguistic variable and its application to
approximate reasoning I, LA Zadeh, 1975
1336 Similarity relations and fuzzy orderings+, LA Zadeh, 1971
1141 Networks of constraints: Fundamental properties and applications
to picture processing, U Montanari, 1974
633 Rough set approach to incomplete information systems, M Kryszkiewicz, 1998
597 A review of fuzzy set aggregation connectives, D Dubois, 1985
561 An introductory survey of fuzzy control, M Sugeno, 1985
513 Fuzzy random variables--I. Definitions and theorems, H Kwakernaak, 1978
475 On the dempster-shafer framework and new combination rules, RR Yager, 1987
448 A procedure for ordering fuzzy subsets of the unit interval, RR
Yager, 1981
438 Rudiments of rough sets, Z Pawlak, 2007
410 Toward a generalized theory of uncertainty (GTU), LA Zadeh, 2005
377 Generalizing database relational algebra for the treatment of
incomplete or uncertain information and vague queries, H Prade, 1984
371 Quantitative fuzzy semantics, LA Zadeh, 1971
340 Rules in incomplete information systems, M Kryszkiewicz, 1999
338 On the analytic formalism of the theory of fuzzy sets, R Bellman, 1973
337 Rough sets and intelligent data analysis, Z Pawlak, 2002

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

Fwd: [bisc-group] Re: Fuzziness vs. Probability: Kolmogorov Axioms

2011-01-04T03:06:00.001-08:00

From: Boris Kovalerchuk Boris.Kovalerchuk@cwu.edu

Dear BISC members,

I am glad that my posts on Kolmogorov probability axioms attracted comments. Some comments are quite emotional (Walenty Ostasiewicz) which show the importance of the topic and differences of opinions.

On the factual side, with all respect I must disagree with comments posted by Walenty that probability P is defined specifically on random events. Accepting this random event position would mean missing a more general view on the probability as a measure and missing the opportunity to build a solid theory for Computing with Words by using it.

Everybody can read Kolmogorov axioms (http://en.wikipedia.org/wiki/Probability_axioms) and see that they have no references to random events. The same can be checked in other languages including Polish (http://pl.wikipedia.org/wiki/Aksjomaty_Ko%C5%82mogorowa). Kolmogorov axioms from the authoritative Encyclopedia of Mathematics (http://eom.springer.de/P/p074970.htm ) and wikipedia are attached.

Kolmogorov described axioms less formally himself for the Encyclopedia of Mathematics also without random events (see below) where the letter omega is substituted by the word omega for readability in plain text.

The mathematical models employed in probability theory are based on three concepts: A space Omega of so-called elementary events, a class of subsets of Omega (events) and a probability distribution — a set function P defined on this class. The value P(A) of P at an event A is then called the probability of A (A.N. Kolmogorov, Probability, Encyclopedia of Mathematics http://eom.springer.de/P/p074890.htm).

It is not accidental that random events are absent in these axioms. To define P on random events one must first define formally random events, which might be difficult to do without the probability concept defined. Therefore, it is done in the opposite way: first, the probability is defined for any event and then a random variable is defined using it. In depth Kolmogorov elaborated the concept of randomness (algorithmic randomness) much later in 1960s (http://en.wikipedia.org/wiki/Kolmogorov_randomness#Kolmogorov_randomness).

Kolmogorov defined the probability as a special additive (sigma-additive) measure on the set of elements of ANY nature that are commonly called events. In statistics and the frequency-based interpretation of probability (common in probability textbooks), they might be called random events, but the general definition of the probability space is not limited by this naming and interpretation as was quoted above. In fact, the set of elements of the probability space must be an algebra (sigma-algebra) of elements of any nature. This important difference is often missed because it is not critical for the traditional use of the probability theory to analyze probability distributions for random events and processes in the frequency-based interpretation of the probability that was a focus for Kolmogorov too.

Why is the general view on probability derived from Kolmogorov axioms important for the BISC community? It allows modeling not only frequency-based probability, but also linguistic uncertainty and subjective probability without violation of Kolmogorov axioms. It is possible because the concept of the probability defined in Kolmogorov axioms does not use the concept of randomness. This idea was exploited in my examples with linguistic uncertainties posted previously including the example of P(Old or Almost Old / John, 59) about wise John.

Traditionally many people automatically associate the word probability with its frequency-based interpretation. In fact in Kolmogorov axioms, P(A) is a generic numeric measure that is named the probability of A. Therefore, more general names such as a measure of uncertainty, a measure of truth, a degree of truth might be applicable to P(A) defined by Kolmogorov axioms.

Conclusion . It would be a mistake to miss this general (measure theory based) view on probability formulated in Kolmogorov axioms and to miss the opportunity to build a solid theory for computing with words by using it.

Regards and Happy Holidays.

Boris

Re: Fuzziness vs. Probability /Gottinger

2011-01-04T03:05:00.001-08:00

From: Paul Snow paulusnix@gmail.com

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Bruno and BISC,

It is interesting that you would cite Shafer's paper in a call for
exploring potential foundations of fuzzy semantics. Much of that paper
is about Cox, but in the context of Jaynes' rewrite of Cox. After all,
Shafer was writing in rebuttal to Van Horn, who had freely attributed
ideas found only in Jaynes to Cox, which was a popular thing to do
once upon a time.

The curiosity is that the possibility calculus, surely something
relevant to fuzzy semantics, is a solution of Cox's axioms, as is now
well-known. This is not only trivially so (since the possibility
calculus can be represented by a class of ordinary probability
distributions), but "substantively" as well. There is no mystery here,
since Cox simply wasn't a Bayesian, _pace_ Jaynes.

As to the Berenstein, if we are to allow counting of cases, then we
may as well adopt Lukasiewicz's 1913 "Logical Probability" (surely
another author whose work might have some relevance to potential
foundations for fuzzy semantics). On the other hand, if we are
exploring how far qualitative axioms will take us when motivating
probabilistic representation of belief, then the reluctant achievement
of Bruno De Finetti stands out.

Poor Bruno, writing under rushed and emotionally charged conditions,
misstated one of his premises, as well as failed to prove what
followed from them. This led to a cottage industry of rehabilitation,
searching out intutive requirements for linear systems expressing
ordinal constraints whose solutions are probability distributions and
only probability distributions.

It turns out that a reconciliation of De Finetti's algebraic notation
with what he says in words saves the day. Some related thinking
provides an "intuitive" meaning for Scott's superficially daunting
main axiom, and so the package is complete, since Scott proved the
required theorem. De Finetti, then, apparently came within a slip of
the pen of proving his case.

There is lots of potential for developing interpretations of fuzzy
calculi analogous to what has been done for probability. For example,
it is well accepted to treat bivalent logic in extensive constant-sum
game terms. The interpretation for AND, OR and NOT goes through easily
for multivalued logic, and if the real unit interval is available,
then there is an asymptotic representation of all bounded linear
programs as fuzzy A-O-N expressions, and _vice versa_.

The hitch, of course, is implication. Since there are many fuzzy
implication operations, and they are generally not expressible in
terms of AND, OR and NOT, we find ourselves cataloging the different
things someone might want to say about the relationships among games
or linear programs, rather than some elegant and meaningful grand
summary.

Still, the potential is there, and fully analogous to what
probabilists have done. They, after all, have never explained why
belief change would be governed by Bayes theorem (as distinct, say,
from how conditional probabilities impart structure to static
beliefs). Or, at least they have not done so in a simple argument
comparable with what is now available for static belief orderings.

Maybe that is a work in progress. Then again, maybe fuzzy is, too.

Happy New Year, everyone.

Paul Snow

[Apart from easily located standard references, this note includes
material presented at greater length in my papers at the 2004 Pacific
Rim AI Conference, p. 445 ff., on games, programs and fuzzy
propositions; the 2005 SIPTA conference, p. 332 ff., and the 2006 IPMU
conference, p. 1993 ff, on the De Finetti &Scott saga.]

On Fri, Dec 17, 2010 at 8:35 PM, Boris Kovalerchuk
<Boris.Kovalerchuk@cwu.edu> wrote:
> *********************************************************************
> Berkeley Initiative in Soft Computing (BISC)
> *********************************************************************
>
> Dear BISC members,
>
> In response to my post on Nov 30, Hans Gottinger raised important issues.
>
> HG: Why not returning to pre-Kolmogorov probability theory.Here <we> have a set of qualitative (binary) relations, which could adopt linguistic terms, identifying qualitative linguistic probability as an axiom system that could be mapped into compatible probability measures.
>
> BK: At the best of my knowledge, this idea was underdeveloped at the pre-Kolmogorov time for linguistic terms. I would appreciate references that I may miss. Glenn Shafer wrote an inspiring analysis of Bernshtein's axioms of probability
>
> (www.glennshafer.com/assets/downloads/other14.pdf):
>
> "In his 1917 article, "On the axiomatic foundation of the theory of probability", Bernstein accepted equally likely cases as the starting point. But instead of arbitrarily defining numerical probability as the number of favorable cases to the total number of cases, he derived this definition from qualitative axioms. Here are his two most important axioms:
>
> . If A and A1 are equally likely, B and B1 are equally likely, A and B are incompatible, and A1 and B1 are incompatible, then (A or B) and (A1 or B1) are equally likely.
>
> . If A occurs, the new probability of a particular occurrence alpha of A is a function of the initial probabilities of alpha and A.
>
> As Shafer emphasized, the important point here is starting from qualitative axioms, not numbers. In the representative measurement theory (Suppes at al, 1966-1990) the numbers are derived from axioms. The same can be beneficial for fuzzy logic.
>
> ...

probability vs fuzziness/ Zadeh

2010-12-14T23:51:00.001-08:00

From: Boris Kovalerchuk <Boris.Kovalerchuk@cwu.edu>

Dear Lotfi,

In your responses, you asked for probabilistic solutions. Below I use the concept of exact complete linguistic context space (references below) based on your concept of the linguistic variable. It is simpler than random sets and satisfies Kolmogorov axioms for probability.

LZ: How would random set theory apply to the following problem? Usually it takes Robert about an hour to get home from work. Usually Robert leaves office at about 5 pm. What is the probability that Robert is home at 6:15 pm?

BK: Consider two linguistic terms about an hour and about two hours. Next we build a probability space S1 with two elementary events S1={about an hour, about two hours} and with conditional probabilities: P(about an hour/1 h 15 min)=0.75, P(about two hours/1 h 15 min)=0.25.

Time 1 h 15 min is selected using 5 pm, and 6.15 pm in your example.

Next we build another probability space S2 with two elementary events S2={usually, rarely} and conditional probabilities, P(usually/1 h 15 min)=0.8, P(rarely/1 h 15 min)=0.2.

For independent events in S1 and S2

P(usually about an hour/1 h 15 min)= P(about an hour/1 h 15 min)* P(usually/1 h 15 min)=0.75*0.8=0.6.

Thus, probability that Robert is home at 6:15 pm is 0.6. For dependent events in S1 and S2 we use conditional probabilities.

LZ: To illustrate, if I am asked to mark on a scale from 0 to 1 the degree to which I like my job, I can do so with ease. How would random set theory apply to this question and my answer? No conditional probability, no voting model and no random sets.

BK: Let m(like job, Lotfi)=0.9 and m(does not like job, Lotfi)=1-09=0.1

In Loginov's interpretation of m as a conditional probability, we get the probability space

{does not like job, like job} with P(like job/ Lotfi)=0.9 and P(does not like job/ Lotfi)=0.1.

In November, I posted the example to estimate how wise is John, 59 knowing that old or almost old people are quite wise. In that example, the probabilistic operation (summation) was better than the fuzzy logic min operation to estimate m(Almost old OR Old, 59). You did not object to this conclusion and commented that in comparing two different methodologies, it is rarely the case that we can assert that A is better than B or B is better than A. Then you added that for some examples A is better than B and for others B is better than A.

Conclusion.

I believe that if we build a collection of examples where both probabilistic and fuzzy logic formulations exist and we agree that A is better than B or B is better than A, then we will be able to teach students by examples very specifically about advantages of A or B.

I agree with you, such examples are absent in probability textbooks, but examples above satisfy Kolmogorov axioms for probability. The focus of my posts was not how to formulate tasks in probabilistic terms vs. fuzzy logic terms, but on finding appropriate operations in the specific examples when both formulations are possible. The values of membership functions and probabilities can be the same, but operations make the difference if John is wise.

Kovalerchuk, B., Klir, G., Linguistic context spaces and modal logic for approximate reasoning and fuzzy probability comparison. In: Proc. of Third International Symposium on Uncertainty Modeling and Analysis and NAFIPS' 95, College Park, Maryland, 1995, IEEE Press, 1995, pp.A23 A28.

Kovalerchuk, B. Context spaces as necessary frames for correct approximate reasoning. International Journal of General Systems, v.25, n 1, 1996, pp. 61-80.

With best regards

Boris

Boris Kovalerchuk, Professor
Dept. of Computer Science,
Central Washington University,
Ellensburg, WA, 98926, USA
Phone: (509) 963-1438;
Fax: (509) 963-1449.
borisk@cwu.edu
http://www.cwu.edu/~borisk

>>> On 11/30/2010 at 5:37 PM, in message <4CF5A6F1.9010107@eecs.berkeley.edu>, "Lotfi A. Zadeh" <zadeh@eecs.berkeley.edu> wrote:

Dear Boris:

    In comparing two different methodologies, it is rarely the case that we can assert that A is better then B or B is better than A. Examples. Differential equations vs. finite-state machines. Evolutionary computing vs. neurocomputing. For some examples A is better than B and for others B is better than A. For this reason, instead of arguing over whether A is better than B or vice versa, it is best to form a partnership of A and B. This is the key idea which underlies soft computing. Soft computing is a partnership of fuzzy logic, neurocomputing, evolutionary computing and probability theory. Note that I have never taken the position that fuzzy logic is better than probability theory. The two are complementary rather than competitive. When we combine the two, resulting in the generalized theory of uncertainty (GTU) (Zadeh 2006), we get a theory which is more general than standard probability theory and hence is more powerful. In an earlier message, I posed a few problems as tests. One of them is the following. X is a real-valued random variable. Usually X is much larger than approximately a. Usually, X is much smaller than approximately b. What is the probability that X is approximately b? How would you apply standard probability theory or random set theory (set-valued probability theory), or Kolmogorov's axioms to this problem. Here is another problem, which is a generalization of a standard problem in probability theory. A box contains black and white balls. I do not know how many. I draw a ball at random and put it back in the box. After a while, I form a perception that most of the time the ball which I draw is white. What is the probability that in n draws of the ball most are white? Did you ever see problems of this kind in textbooks on probability theory? How would you apply Kolmogorov's axioms to this problem? What should be noted is that in this and the earlier problem, the boundaries of words are unsharp. Fuzzy logic is not a logic of randomness or likelihood. It is a logic of classes with unsharp boundaries. However, what should be stressed is that concepts drawn from probability theory has always played an important role in fuzzy logic.

    I await your solutions. Will anybody else volunteer to come up with solutions based on standard probability theory or random set theory.

        With my warm regards.

        Sincerely,

        Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

What is different between Partiality and Granularity?

2010-12-07T23:49:00.001-08:00

Dear All BISC Members,

In the view of Prof.Zadeh PNL is trying to achieve two human
capabilities for describing perceptions: Partiality and Granularity.
So partiality tell us that many of human concepts are a matter of
degree. Concepts like: relevance, truth, likelihood, . . .
For example if P is : "This subject is very relevant to our work" then
it would be interpreted like this: Degree(relevance(subject, our work)
is high).
Consequently we have granulated Degree of relevance not further more !
so partiality is not more than Degree granularity?

All the best,

------------------------------------------------------
Mehdi Rohaninezhad
UKM - National University of Malaysia
School of Computer Science
Call : +60133564061

self-introduction

2010-12-07T23:49:00.000-08:00

From: <Michael.Mlynski@arcor.de>

Dear Lotfi and members of the BISC Group,
thanks a lot for your invitation to self-introduce myself to the BISC group.
My name is Michael Mlynski, senior software engineer and project
manager at EUtech Scientific Engineering GmbH, a rather small company
in Aachen, Germany. I obtained my BA in electrical and electronic
engineering at the Technical University Karlsruhe in 1995 and my MA in
communications engineering at the Ruhr-University Bochum in 1998, both
in Germany. I send spend a couple of years in industry as a hardware
and later software engineer. In 2003 I obtained my PhD from the
Technical University Aachen, Germany.
My research interest are methods to represent as well as apply
imprecise statements and knowledge. In detail I focus on methods which
can be easily and effectively implemented in software. Verification,
adjustment and maintenance are just a few of many topics in this
fields and apply to both, the software as well as the knowledge being
implemented. Besides many papers I have written a major article
together with Hans-Jürgen Zimmermann introducing a mathematic
framework which allows application-oriented knowledge representation
and processing.
As an engineer in industry my time for research work is very limited.
Periodical I have to put the majority of my ideas on ice,
concentrating on rather small research projects which I'm capable to
realize.
I'm in my 40s and have a wife, a daughter and two sons. We live in
Aachen, Germany, which is located just north of the "Eifel", a low
mountain range where I love to go free climbing. And yes, besides my
job, my research activities, my family and free climbing, once in a
while there is some time left to draw cartoons (there also exists a
short cartoon of mine about computer viruses on YouTube).
With regards
Michael Mlynski
-----------------------------------------------
Dr.-Ing. Michael Mlynski
Am Bollet 57
D-52078 Aachen
Germany
Phone: +49 / (0)241 / 52 72 96
E-Mail: Michael.Mlynski@arcor.de

Pictures are louder than words

2010-12-07T23:48:00.002-08:00

From: Paul P. Wang <ppw@ee.duke.edu>

Dear Larry:

What you are saying is quite accurate.This is a very complex issue,the
creativity has a great deal to do with the growth,cultural
evolution,political leadership,the creative environment,living
condition,economics and the motivation factors.Before Royal Swedish
Academy of Sciences and Nobel time,there were no lack of crucial and
significant inventions ,both theoretical and pragmatical,contributed
by native Chinese great thinkers.Imagine the world without paper
money,compass,silk,.....or Tofu?

Best

Paul

Amusing Aphorisms

2010-12-07T23:48:00.001-08:00

From: John Meech <jameech@dccnet.com>

Knowing how much Professor Zadeh appreciates aphorisms,
I recently came across these two quotations about uncertainty
and felt I should pass them on to the BISC Group:

"I can live with doubt and uncertainty and not knowing.
I have approximate answers and possible beliefs and different
degrees of certainty about different things...it doesn't frighten me."
- Richard Feynman

"There must be certainty from the President of the United States."
- George W. Bush

Best wishes to all,

John

John A. Meech, P.Eng., Ph.D.
Professor of Mining Engineering and Director of CERM3
(The Centre for Environmental Research in Minerals, Metals, and Materials),
The University of British Columbia,
The Norman B. Keevil Institute of Mining Engineering,
6350 Stores Road, Vancouver, V6T 1Z4, CANADA

Email: jameech@dccnet.com
Tel: 604-822-3984
Fax: 604-822-5599
Cell: 604-761-0472

"There is no failure...only early attempts at success. " - Mike Myers
(Guru Pitka)

[bisc-group] Re: Fuzziness vs. Probability: Ostasiewicz

2010-12-07T23:47:00.001-08:00

From: Boris Kovalerchuk <Boris.Kovalerchuk@cwu.edu>

Dear Walenty,

I guess that you missed my points. Below I clarify them.

WO: From Boris: m(tall,170)=0.45 is not an random event,

BK: I have no m(tall,170)=0.45 in my example, but I have
m(old,59)=0.45 taken from fuzzy literature (see reference in my
previous post). I did not say if it is a random or not random event.

WO: … probability is a function of an argument is a random event, at
least in the most of known and practiced probability theories.

BK: You likely missed my original message (posted on Nov 5, 2010:)
"Next, we have at least three flavors of probability: (1)
frequency-based, (2) subjective, and (2) axiomatic (Kolmogorov's
axioms). The last one abstracts the first two interpretations and
maybe others yet unknown. Thus, thinking about probabilities only as
frequencies of repeating events is a very narrow view of probability
that should be avoided when we try to distinguish fuzzy membership
functions from probabilities."

Kolmogorov's axioms of probability (probability space) are formulated
without random events
(http://en.wikipedia.org/wiki/Probability_axioms).

WO: T-norms are probabilistic operations, they were imported to fuzzy
sets theory, existed a long before.

BK: Yes, they existed before. But was the goal changed? In fuzzy logic
T-norms are used to define fuzzy logic AND operations. The probability
theory does not use T-norms to define AND operator (intersection)
within the probability space. In a probabilistic metric space,
T-norms are used to define a generalized distance using probability
distributions, but the probabilistic metric space differs from the
Kolmogorov's probability space
(http://en.wikipedia.org/wiki/Probability_space).

WO: The question" should we use probabilistic or fuzzy logic
operations" is of the same kind as should we use computers or cars?

BK: The question is taken out of its context. The full question was:
"If we produce membership functions in a probabilistic way, why not
use the probabilistic operations to combine them (union and
intersection) and instead use multiple T-norms and T-conorms?

Thus, it is about the use of T-norms and T-conorms instead of specific
probabilistic operations (intersection and union). I provided the
example about "wise John" where T-conorm (max) and the probabilistic
union operation (summation) are used for the same purpose to evaluate
if John is wise. Therefore, my question is not about comparing
"computers and cars". It is about comparing two "computers" that
differently implement OR operation for the expression "old OR almost
old".

Regards

Boris

Free downloads from Logica Universalis

2010-12-02T22:37:00.001-08:00

From: John F. Sowa <sowa@vivomind.com>

The current issue of the journal Logica Universalis contains six
articles, which can be downloaded for free from now to December 31:

http://www.springerlink.com/content/n109547vx662/

The authors of the articles were asked to answer the following
questions:

1. Do all human beings have the same capacity of reasoning? Do men,
women, children, Papuans, yuppies, reason in the same way?

2. Does reasoning evolve? Did human beings reason in the same way two
centuries ago? In the future will human beings reason in the same
way? Are computers changing our way of reasoning? Is a mathematical
proof independent of time and culture?

3. Do we reason in different ways depending on the situation? Do we
use the same logic for everyday life, in physics, and in questions
to do with the economy?

4. Do the different systems of logic reflect the diversity of
reasoning?

5. Is there any absolute true way of reasoning?

John Sowa
______________________________________________________________________

Contents of Logica Universalis, Volume 4, Number 2 / November 2010

Preface: Is Logic Universal?
Jean-Yves Beziau
http://www.springerlink.com/content/dv2q525590475157/

Human Rationality Challenges Universal Logic
Brian R. Gaines
http://www.springerlink.com/content/y7400543luq15pr8/

Logic and Natural Selection
Jaroslav Peregrin
http://www.springerlink.com/content/4673765064677577/

The Place of Logic in Reasoning
Daniel Kayser
http://www.springerlink.com/content/q3852284771m3521/

Is Logic Necessary?
Gregory McColm
http://www.springerlink.com/content/38n0560670460l5t/

Carnap, Goguen, and the Hyperontologies: Logical Pluralism and
Heterogeneous Structuring in Ontology Design
Oliver Kutz, Till Mossakowski and Dominik Lücke
http://www.springerlink.com/content/e170073t6676m765/

What is different between Partiality and Granularity?

2010-12-02T22:36:00.003-08:00

From: mehdi rohaninezhad <mr@ftsm.ukm.my>

Dear All BISC Members,

In the view of Prof.Zadeh PNL is trying to achieve two human capabilities for describing perceptions: Partiality and Granularity.
So partiality tell us that many of human concepts are a matter of degree. Concepts like: relevance, truth, likelihood, . . .
For example if P is : "This subject is very relevant to our work" then it would be interpreted like this: Degree(relevance(subject, our work) is high).
Consequently we have granulated Degree of relevance not further more ! so partiality is not more than Degree granularity?

All the best,

Mehdi Rohaninezhad
UKM - Malaysia

Fuzzy Logic is an approach to thinking that thrives on diversit

2010-12-02T22:36:00.001-08:00

"John Meech" <jameech@dccnet.com>

Fuzzy Logic is an approach to thinking that thrives on diversity. It revels in the fact that when we approach a problem or make a decision, each of us use our diverse past experiences and personal expertise to arrive at our conclusion. It does not demand that the conclusion be optimal (how ever that may be defined). It accepts the fact that eventually, if the principles of knowledge that are being applied are appropriate and valid in the context of the problem space, the best result will emerge. It allows us to modify our knowledge over time as new ideas come into the problem space. It tolerates the fact that our measurements are imprecise and our knowledge is lacking. It is a methodology that can derive specific models or it is one that allows these models to be non-specific and to incorporate biases and generalities. Sometimes these biases and generalizations are weak and so, an open mind is necessary to adapt one's thinking and dispense with one's prejudices.

There are few books on Complex Systems that focus on Fuzzy Logic, but one that tells us about the value of Diversity is the following:

Diversity and Complexity (Primers in Complex Systems) by Scott E Page

I recommend it to those who are seeking more precision in how fuzzy logic works.

Regards to all,

John

John A. Meech, P.Eng., Ph.D.
Professor of Mining Engineering and Director of CERM3
(The Centre for Environmental Research in Minerals, Metals, and Materials),
The University of British Columbia,
The Norman B. Keevil Institute of Mining Engineering,
6350 Stores Road, Vancouver, V6T 1Z4, CANADA

Email: jameech@dccnet.com
Tel: 604-822-3984
Fax: 604-822-5599
Cell: 604-761-0472

[bisc-group] Diversity and Complexity

2010-12-02T00:32:00.001-08:00

From: John Meech <jameech@dccnet.com>

Fuzzy Logic is an approach to thinking that thrives on diversity. It revels in the fact that when we approach a problem or make a decision, each of us use our diverse past experiences and personal expertise to arrive at our conclusion. It does not demand that the conclusion be optimal (how ever that may be defined). It accepts the fact that eventually, if the principles of knowledge that are being applied are appropriate and valid in the context of the problem space, the best result will emerge. It allows us to modify our knowledge over time as new ideas come into the problem space. It tolerates the fact that our measurements are imprecise and our knowledge is lacking. It is a methodology that can derive specific models or it is one that allows these models to be non-specific and to incorporate biases and generalities. Sometimes these biases and generalizations are weak and so, an open mind is necessary to adapt one's thinking and dispense with one's prejudices.

There are few books on Complex Systems that focus on Fuzzy Logic, but one that tells us about the value of Diversity is the following:

Diversity and Complexity (Primers in Complex Systems) by Scott E Page

I recommend it to those who are seeking more precision in how fuzzy logic works.

Regards to all,

John

John A. Meech, P.Eng., Ph.D.
Professor of Mining Engineering and Director of CERM3
(The Centre for Environmental Research in Minerals, Metals, and Materials),
The University of British Columbia,
The Norman B. Keevil Institute of Mining Engineering,
6350 Stores Road, Vancouver, V6T 1Z4, CANADA

Email: jameech@dccnet.com
Tel: 604-822-3984
Fax: 604-822-5599
Cell: 604-761-0472

"There is no failure...only early attempts at success. " - Mike Myers (Guru Pitka)

Reply from Professor ZADEH

2010-12-01T00:23:00.001-08:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>
Subject: [bisc-group] probability vs fuzziness/ Kovalerchuk
To: Boris Kovalerchuk <Boris.Kovalerchuk@cwu.edu>

Dear Boris:

In comparing two different methodologies, it is rarely the case
that we can assert that A is better then B or B is better than A.
Examples. Differential equations vs. finite-state machines.
Evolutionary computing vs. neurocomputing. For some examples A is
better than B and for others B is better than A. For this reason,
instead of arguing over whether A is better than B or vice versa, it
is best to form a partnership of A and B. This is the key idea which
underlies soft computing. Soft computing is a partnership of fuzzy
logic, neurocomputing, evolutionary computing and probability theory.
Note that I have never taken the position that fuzzy logic is better
than probability theory. The two are complementary rather than
competitive. When we combine the two, resulting in the generalized
theory of uncertainty (GTU) (Zadeh 2006), we get a theory which is
more general than standard probability theory and hence is more
powerful. In an earlier message, I posed a few problems as tests. One
of them is the following. X is a real-valued random variable. Usually
X is much larger than approximately a. Usually, X is much smaller than
approximately b. What is the probability that X is approximately b?
How would you apply standard probability theory or random set theory
(set-valued probability theory), or Kolmogorov's axioms to this
problem. Here is another problem, which is a generalization of a
standard problem in probability theory. A box contains black and white
balls. I do not know how many. I draw a ball at random and put it back
in the box. After a while, I form a perception that most of the time
the ball which I draw is white. What is the probability that in n
draws of the ball most are white? Did you ever see problems of this
kind in textbooks on probability theory? How would you apply
Kolmogorov's axioms to this problem? What should be noted is that in
this and the earlier problem, the boundaries of words are unsharp.
Fuzzy logic is not a logic of randomness or likelihood. It is a logic
of classes with unsharp boundaries. However, what should be stressed
is that concepts drawn from probability theory has always played an
important role in fuzzy logic.

I await your solutions. Will anybody else volunteer to come up
with solutions based on standard probability theory or random set
theory.

With my warm regards.

Sincerely,

Lotfi

--
Lotfi A. Zadeh
Professor in the Graduate School
Director, Berkeley Initiative in Soft Computing (BISC)

BISC Homepage URLs
URL: http://zadeh.cs.berkeley.edu/

Re: Fuzziness vs. Probability

2010-12-01T00:22:00.001-08:00

From: Boris Kovalerchuk <Boris.Kovalerchuk@cwu.edu>

Lotfi suggested to generalize the probability theory by adding
linguistic concepts from fuzzy logic (1995, 2009). This is a very
valid point. However, the devil is in the details. Should we use
probabilistic or fuzzy logic operations? I asked in my previous post:
If we produce membership functions in a probabilistic way, why not use
the probabilistic operations to combine them (union and intersection)
and instead use multiple T-norms and T-conorms?

The answer will also clarify the differences between fuzzy logic and
probability more specifically than comments that do not touch the
operations. Vladik presented random sets as an answer with
probabilistic operations. Lotfi presented some general concerns about
applicability of random sets. An alternative linguistic context space
(LCS) approach (Kovalerchuk, Klir, 1995; Kovalerchuk, 1996, 2000) in
shown below to answer this question based on Lotfi's concept of
linguistic variable.

Example. A knowledge base contains a statement: "Old or almost old
people are quite wise." The reasoning model is to conclude that John
is quite wise if he is old or almost old. What is the probability that
John, 59 is quite wise?

What is the fuzzy membership function value that John, 59 is quite wise?

Let membership values be m(old,59)=0.45, and m(almost old,59)=0.55.
These numbers are close to values from the frequency-based experiment
(Hall, Kandel, Szabo 1986). Then m(old OR almost old, 59)=max
(0.45,0,55)=0.55 using the fuzzy logic max operator for OR. In
essence, 0.55 means a refusal to judge how wise John is.

Alternatively, let conditional probabilities be P(old/59)=0.45, and
P(almost old/59)=0.55, then

P(old OR almost old/59)= P(old/59)+ P(almost old/59)=0.45+0.55=1. This
is a certain answer that John, 59 is quite wise. Which of two answers
fits better to our common sense reasoning? If we vote I thing that

m(old or almost old, 59)=max (0.45,0,55)=0.55 will not win over P(old
OR almost old/59)=1 .

The answer P(old OR almost old/59)= 1 is based on the probability
space with two uncertain linguistic labels as elementary events {old,
almost old} with probabilities that satisfy Kolmogorov's axioms. We
call it the exact complete Linguistic Context Space for age 59. For
age 39, it could be a different space {young, middle age, almost old}.
This approach differs from Loginov's conditional probability
interpretation (Zadeh, 1995) where we have two probability spaces and
sets of elementary events: S1={old, not old} and S2={almost old, not
almost old} with conditional probabilities P1(old/59), P2(almost
old/59) from two different probability spaces that should not be
added. Adding them would violate probability laws. For some ages,
natural language provides incomplete or overcomplete sets of
linguistic terms with sums less or greater than 1. In "engineered"
natural languages used in fuzzy control, the exact linguistic context
spaces are often made, which is an advantage of fuzzy control and its
link with the probability theory.

This example raises the question: "Do we have arguments to claim a
monopoly for reasoning with linguistic concepts with unsharp
boundaries based on T-norms and T-conorms fuzzy logic operations?"

Conclusion.

We need to distinguish: (1) linguistic fuzzy logic (LFL) and (2)
T-Computational Fuzzy Logic (TCFL) based on T-norms and T-conorms. In
the comparisons with probability, most of the arguments for fuzzy
logic are concentrated in (1), and most of the arguments against are
concentrated in (2). Thus, two sides often talk about different
things including our current discussion. The clarity of terminology
may avoid this. There is agreement about advantages of (1), thus it
makes sense to concentrate the discussion on (2) with less agreement.

Regards

Boris

What is the difference between ISu and ISbm in PNL?

2010-11-26T00:56:00.001-08:00

From: mehdi rohaninezhad <rohaninezhad@gmail.com>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear All BISC members,

I suppose that all followed propositions are a combination of possibility and probability.
1- X ISp R    : X is random variable
2- X ISu R    : X is random variable
3- X ISbm R : (type1) X is real valued random variable , (type 2) X is fuzzy set valued random variable
4- X ISrs R   : X is fuzzy set valued random variable

As far as I've understood ISp is a combination of crisp probability with possibility, But ISu and ISbm are combination of granular probability and possibility.
For example:
ISp -> 0.2\small+0.7\medium+0.1\large
ISU,ISbm -> low\small+high\medium+low\

large

My question is:

1- what is the difference between ISu and ISbm?
For example, [Prob{X is small} is usually] can be assigned to both of propositions: ISu, ISbm ?

2- what is the difference between ISbm type 1 and type 2?
For example: type 1: X ISbm (low\2+high\7+low\10) ?
type 2: X ISbm (low\small+high\medium+low\large) ?
Please give me a linguistic example.

Warm Regards,

--
Mehdi Rohaninezhad
Call : +60133564061

[bisc-group] UNCERTAINTY

2010-11-26T00:54:00.001-08:00

In social, legal and above all logical-scientific reasoning, what we
need, first and foremost, is having *clear concepts*. Fuzzy-experts
and kindred professionals should not trifle with this. Talking about
fuzziness, one should not be fuzzy. Down with mystifications!

I agree that clarity is a goal in science, but it's always the end
point, not the starting point of research.

I also agree that it's important to formulate clear, testable
hypotheses that can be falsified by well-designed experiments.
But even that is not easy, especially in the early stages of
any field -- science, engineering, philosophy, or life.

And I certainly agree that deliberate mystification is bad, but
the opposite -- pretending that something is clear -- can be
equally bad. Following is a quotation by Clarence Irving Lewis
about the state of the art of philosophy in 1960:

CIL:

It is so easy... to get impressive "results" by replacing the
vaguer concepts which convey real meaning by virtue of common
usage by pseudo precise concepts which are manipulable by
"exact" methods -- the trouble being that nobody any longer knows
whether anything actual or of practical import is being discussed.

Another quotation, attributed to Lord Kelvin, summarizes the issue:

LK:

Better a rough answer to the right question
than an exact answer to the wrong question.

From: John Meech [bisc-group] UNCERTAINTY

2010-11-24T01:58:00.001-08:00

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Hans Kuijper's statement "Talking about fuzziness, one should not be fuzzy" forces me to respond.

I object to this statement, although I do understand Dr. Keijper's frustration with the incredible range of terminology that overlaps in this field.

He should ask himself why we have so many different languages and cultures in the world? What is the point of so much diversity - surely we should have a single common language and we should all think the same way. What is beautiful to one person should not be handsome to another. There is no need for such a myriad of hedging terms - very, almost, exactly, virtually, nearly, close to, definitely, more or less, and so on. But if we do decide to standardize these terms, then everyone must use the agreed-upon definition.

The world is fuzzy - live with it! - deal with it! It is fuzzy because each of us see things in different ways, different contexts, and with different knowledge. And knowledge is never complete, so what is right today might be wrong tomorrow because of changes in the environment or design or context or amount of information.

Zadeh recognized this over 55 years ago - why is it taking the world so long to realize that fuzziness is a description of the uncertainty that surrounds us in everything we do or think about? And the type of fuzziness can also change in itself dependent on beliefs, culture, past experiences, and training.

By the way "not being fuzzy" is a complementary function to "being fuzzy", so even when Dr. Kuijper states that he wants to be "not fuzzy", unfortunately he really is "being fuzzy"!

So be it.

Best wishes to all,

John

John A. Meech, P.Eng., Ph.D.
Professor of Mining Engineering and Director of CERM3
(The Centre for Environmental Research in Minerals, Metals, and Materials),
The University of British Columbia,
The Norman B. Keevil Institute of Mining Engineering,
6350 Stores Road, Vancouver, V6T 1Z4, CANADA

Email: jameech@dccnet.com
Tel: 604-822-3984
Fax: 604-822-5599
Cell: 604-761-0472

"There is no failure...only early attempts at success. " - Mike Myers (Guru Pitka)

On Sat, Nov 20, 2010 at 2:21 PM, hans kuijper <j_kuijper@online.nl> wrote:

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear BISC members,

On October 26, I reminded you of the email I had sent on 13-12-2009 (Re: UNCERTAINTY), an email apparently none of you has been able to respond to satisfactorily. Afterwards, I have been lucky enough to receive from you different kinds of messages, some of them accompanied by surprising photographs. However, I'm still wondering a) what the differences and similarities are between 'approximate sets', 'credibility sets', 'fuzzy sets', 'grey sets', 'imprecise sets', 'possibility sets', 'random sets', 'risk sets', 'rough sets' and 'vague sets', b) how these sets (often varied in themselves!) are related to that bewildering array of 'probabilities', and c) in what way all this has to do with 'information', the elusive concept that T.S. Eliot had in mind when he wrote:

'Where is the wisdom we have lost in knowledge?

Where is the knowledge we have lost in information?'

Next year, in September, the 2011 IEEE International Conference on Grey Systems and Intelligent Services will be held in Nanjing, not far from the place where my hero, the genius Wang Bi, lived his short life (http://ieeegsis.nuaa.edu.cn). It seems to me that the conference - at which Professor Julong Dong, founder of the blooming and flowering grey systems theory ('grey' being a mixture, or blend, of black and white!), is scheduled to be one of the keynote speakers - will be for experts in the theories of (a posteriori, a priori, aleatory, approximate, Bayesian, binomial, classical, comparative, conditional, counterfactual, empirical, epistemic, experimental, exotic, frequency, game theoretic, geometric, imprecise, inductive, interval, inverse, logical, measure theoretic, negative, objective, pignistic, positive, propensity, qualitative, quantitative, quantum, or subjective) probability, credibility, possibility, stochasticity, fuzzy systems, rough systems, imprecision, approximation, vagueness, randomness, chances, risks, and games a window of opportunity to go beyond the boundaries of their different and often competing disciplines.

In social, legal and above all logical-scientific reasoning, what we need, first and foremost, is having clear concepts. Fuzzy-experts and kindred professionals should not trifle with this. Talking about fuzziness, one should not be fuzzy. Down with mystifications!

Best wishes,

Hans Kuijper

(sinologist turned systemicist)

Rotterdam

UNCERTAINTY

2010-11-22T23:47:00.001-08:00

From: hans kuijper <j_kuijper@online.nl>

Dear BISC members,

On October 26, I reminded you of the email I had sent on 13-12-2009 (Re: UNCERTAINTY), an email apparently none of you has been able to respond to satisfactorily. Afterwards, I have been lucky enough to receive from you different kinds of messages, some of them accompanied by surprising photographs. However, I'm still wondering a) what the differences and similarities are between 'approximate sets', 'credibility sets', 'fuzzy sets', 'grey sets', 'imprecise sets', 'possibility sets', 'random sets', 'risk sets', 'rough sets' and 'vague sets', b) how these sets (often varied in themselves!) are related to that bewildering array of 'probabilities', and c) in what way all this has to do with 'information', the elusive concept that T.S. Eliot had in mind when he wrote:

'Where is the wisdom we have lost in knowledge?

Where is the knowledge we have lost in information?'

Next year, in September, the 2011 IEEE International Conference on Grey Systems and Intelligent Services will be held in Nanjing, not far from the place where my hero, the genius Wang Bi, lived his short life (http://ieeegsis.nuaa.edu.cn). It seems to me that the conference - at which Professor Julong Dong, founder of the blooming and flowering grey systems theory ('grey' being a mixture, or blend, of black and white!), is scheduled to be one of the keynote speakers - will be for experts in the theories of (a posteriori, a priori, aleatory, approximate, Bayesian, binomial, classical, comparative, conditional, counterfactual, empirical, epistemic, experimental, exotic, frequency, game theoretic, geometric, imprecise, inductive, interval, inverse, logical, measure theoretic, negative, objective, pignistic, positive, propensity, qualitative, quantitative, quantum, or subjective) probability, credibility, possibility, stochasticity, fuzzy systems, rough systems, imprecision, approximation, vagueness, randomness, chances, risks, and games a window of opportunity to go beyond the boundaries of their different and often competing disciplines.

In social, legal and above all logical-scientific reasoning, what we need, first and foremost, is having clear concepts. Fuzzy-experts and kindred professionals should not trifle with this. Talking about fuzziness, one should not be fuzzy. Down with mystifications!

Best wishes,

Hans Kuijper

(sinologist turned systemicist)

Rotterdam

Partiality and granularity

2010-11-19T01:32:00.001-08:00

From: mehdi rohaninezhad <mr@ftsm.ukm.my>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear All BISC members,

PNL is trying to achieve two human capabilities in perception descriptions: partiality and granularity.
For example :
1- Prob{X is small} is usually
2- Jun usually drink alcohol

Usually in these propositions are a matter of partiality or granularity?

Warm Regards,

What is difference between ISbm and ISp and ISu in canonical form of PNL?

2010-11-19T01:31:00.001-08:00

Dear All BISC members,

I suppose that all followed propositions are a combination of possibility and probability.
1- X ISp R    : X is random variable
2- X ISu R    : X is random variable
3- X ISbm R : (type1) X is real valued random variable , (type 2) X is fuzzy set valued random variable
4- X ISrs R   : X is fuzzy set valued random variable

As far as I've understood ISp is a combination of crisp probability with possibility, But ISu and ISbm are combination of granular probability and possibility.
For example:
ISp -> 0.2\small+0.7\medium+0.1\large
ISU,ISbm -> low\small+high\medium+low\large

My question is:

1- what is the difference between ISu and ISbm?
For example, [Prob{X is small} is usually] can be assigned to both of propositions: ISu, ISbm ?

2- what is the difference between ISbm type 1 and type 2?
For example: type 1: X ISbm (low\2+high\7+low\10) ?
                    type 2: X ISbm (low\small+high\medium+low\large) ?
Please give me a linguistic example.

Warm Regards,

There are four possibilities of making a statement:

2010-11-18T00:18:00.001-08:00

> There are four possibilities of making a statement:
>
> I) X is F (affirmation),
>
> II) X is not-F (negation),
>
> III) X is both F and not-F (statements in this category can be subdivided
> into 'F to the same degree as not-F', 'more F than not-F' and 'less F than
> not-F'),
>
> IV) X is neither F nor not-F.
>

Please note that the four possibilities has no wuji, no taiji, no Yin
and no Yang. They were devised when

(1) the ancient Greeks rejected negative numbers as absurd;

(2) Aristotle had no way to see the existence and difference of
electron and positron;

(3) the Earth was considered the center of the universe;

> I surmise that the author(s) of the Yijing (and of the Daodejing), who
> said that Yin and Yang (and their intricate interactions) have spontaneously
> emerged from the Great Ultimate (Taiji), which has mysteriously come from
> the Unlimited or Infinite (Wuji), had this tetralemma in mind, for which
> reason it didn't occur to them to introduce the law of excluded middle
> (tertium non datur), and the placing of white/right/hot etc over against
> black/wrong/cold etc was strange to them. Generally speaking, in the face of
> constant change and shifting actuality, the Chinese are more inclined to
> lumping/combining than to splitting/separating.

It seems to me I already have the minimal logical representations:

(0,0) to (-infinite,+infinite): the Unlimited or Infinite (Wuji);

(-1,0) or (-x,0): the yin in crisp or fuzzy value;

(0,+1) or (0,+x): the yang in crisp or fuzzy value;

(-1,+1) or (-x,+x): the Eastern lumping/combining Great Ultimate
(Taiji) in crisp or fuzzy value.

The crisp values leads to bipolar DeMorgan's laws, excluded middle and
thus the fusion of Boolean logic into the YinYang logic.
The fuzzy values leads to fuzzy bipolar DeMorgan's laws, violation of
excluded middle and thus the fusion of fuzzy logic to the YinYang
logic.
The bipolar universal modus ponens (BUMP) leads to bipolar quantum
entanglement.
The three aspects are all in one.

Hope the above can bring some hope to the direction you are pursuing.

Best Regards.
Wen-Ran

Fuzziness vs. Probability Dear Nikola, Boris, Ron and Vladik,

2010-11-09T01:47:00.003-08:00

From: Lotfi A. Zadeh <zadeh@eecs.berkeley.edu>
Date: 9 November 2010 04:18
Subject: [bisc-group] Fuzziness vs. Probability
To: bisc-group@lists.eecs.berkeley.edu
Cc: Jellat Arkadia <jellat007@hotmail.com>, Boris Kovalerchuk <Boris.Kovalerchuk@cwu.edu>, Vladik Kreinovich <vladik@cs.utep.edu>, "Ronald P. Loui" <r.p.loui@gmail.com>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Nikola, Boris, Ron and Vladik,

    The questions posed in your messages have been objects of discussion and debate since the debut of fuzzy set theory. In a 1966 paper, Loginov suggested that a membership function may be interpreted as a conditional probability. In a 1995 paper entitled "Probability Theory and Fuzzy Logic are Complementary Rather Than Competitive," I discussed various connections between probability theory and fuzzy set theory, and pointed out that the conditional probability interpretation, the voting model interpretation, and the random set interpretation are equivalent. In large measure, the issues raised in your messages were addressed in that paper.

    There are points of contact between probability theory and fuzzy set theory, but the theories are distinct, as are the concepts of fuzziness and randomness. Fuzzy set theory is a theory of classes with unsharp boundaries. Probability theory is a formalization of perceptions of likelihood. Fuzzy set theory exploits the remarkable human capability to graduate perceptions, that is, associate perceptions with degrees. To illustrate, if I am asked to mark on a scale from 0 to 1 the degree to which I like my job, I can do so with ease. How would random set theory apply to this question and my answer? No conditional probability, no voting model and no random sets. Another example. Suppose that I am asked to mark on a scale the degree to which Robert is bald. The same applies to this example. The same applies to the Richter scale. The same would apply to the definition of recession. What is involved in these and other examples is the fact that in most of the applications of fuzzy logic membership functions are declared or are in the mind of the utterer, rather than elicited. An example is the way in which fuzzy sets are defined in the rule-base of Honda fuzzy logic transmission. See attachment. In this respect, membership functions are radically different from probability distributions. Suppose that I am asked "What is the probability that Obama will be reelected?" Suppose that my answer is "not high." How would random set theory apply? How would Kolmogorov's axioms apply? How would random set theory apply to the following problem? Usually it takes Robert about an hour to get home from work. Usually Robert leaves office at about 5 pm. What is the probability that Robert is home at 6:15 pm? Another problem. X is a real-valued random variable. What we know about X is that usually X is much larger than approximately a, and usually X is much smaller than approximately b. What is the probability that X is approximately c, where c is a number between a and b? I can construct example after example in this spirit--examples which do not lend themselves to solutions by probability theory.

    Note that conceptually fuzzy set theory is much simpler than probability theory. Theory of random sets is a theory of set-valued random variables. To understand the theory of set-valued random variables requires a Ph.D. in probability theory or mathematics. How would Kolmogorov's axioms apply to set-valued, fuzzy-set-valued random variables and fuzzy events?

    Fuzziness is a pervasive phenomenon in the real-world. To enhance the ability of probability theory to deal with real-world problems, it is expedient to generalize probability theory through addition of concepts and techniques drawn from fuzzy logic. This is what is done in my 2002, 2005 and 2006 papers.

    A concept which plays an important role in generalized probability theory is that of possibility. Possibility theory is distinct from probability theory. In many real-world problems what we encounter are possibility distributions of probability distributions.

    If you are interested in connections between probability theory and fuzzy set theory, you may find it worthwhile to take a look at my 1995 paper and my more recent papers on the generalized theory of uncertainty.

        Regards to all.

        Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

Question from an undergraduate student

2010-11-09T01:47:00.001-08:00

From: <hchongfu@bnu.edu.cn>
Date: 2010/11/6
Subject: Re:[bisc-group] RE: Question from an undergraduate student
To: bisc-group@lists.eecs.berkeley.edu
Cc: Boris.Kovalerchuk@cwu.edu

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Prof.Boris,

To simply the problem, it seems that, a fuzzy set is a bit like a Chinese word, but a probability a number. We use Chinese to write a paper for a journal, but data to calculate.

Regards

Chongfu Huang
Prof. Beijing Normal University
http://adrem.org.cn/Faculty/HuangCF/English/ehcf.htm

------------------ 原始邮件 ------------------
>From: Boris Kovalerchuk <Boris.Kovalerchuk@cwu.EDU>
>Reply-To: bisc-group@lists.eecs.berkeley.edu
>To: "bisc-group@lists.eecs.berkeley.edu" <bisc-group@lists.eecs.berkeley.edu>
>Subject: Re:[bisc-group] RE: Question from an undergraduate student
>Date:Thu, 04 Nov 2010 22:18:35 -0700
>

Dear BISC members,

Let me play a devil's advocate relative to the undergraduate student's questions on relations between fuzzy sets and probability. First I would rather agree with Hans that answers on differences in the "zoo" of uncertainties could be better.

There is a popular idea that "Old" and "Young" are imprecise overlapping concepts and therefore need to be modeled by using fuzzy set membership functions. ("Such natural language terms are used by experts all the time, so to describe them, we need fuzzy logic." V. Kreinovich, BISC, 10/25/2010).

It is commonly contrasted with the probability theory, which deals with crisp disjoint (mutually exclusive) elementary events, e.g., die sides. This was stated as a major difference between the fuzzy set theory and the probability theory. "Logic and probability assume "crisp" predicates and classical set-theoretic semantics" (R. Loui, BISC, 11/2/2010).

This answer is incomplete and leads us to the conceptual difficulties. As often the case, the devil is in the details. This answer did not tell us how to construct these Membership Functions (MFs).
Nikola asked that question recently. Vladik Kreinovich answered that one of the ways is using frequencies of subjective human answers (".we can poll several experts, and if 6 out of 8 says that the guy is tall, we take 6/8 = 0.75."). The fuzzy literature is full of such frequencies for computing MFs for ages Young, Old, Middle Age etc, e.g., Hall, Kandel, Szabo,1986.

If we construct MFs by using such frequencies then it seems that we start to contradict to ourselves.
We denied probabilistic approach for modeling overlapping imprecise linguistic terms and then use it via frequencies to build fuzzy membership functions for these terms.

Now about the next devil in this zoo. If we produce MFs in a probabilistic way (frequency-based or subjective), why do we not use the probabilistic operations to combine them (union and intersection) but use multiple T-norms and T-conorms?

Next, we have at least three flavors of probability: (1) frequency-based, (2) subjective, and (2) axiomatic (Kolmogorov's axioms). The last one abstracts the first two interpretations and maybe others yet unknown. Thus, thinking about probabilities only as frequencies of repeating events is a very narrow view of probability that should be avoided when we try to distinguish fuzzy membership functions from probabilities.

Comments welcome.

Regards

Boris

Boris Kovalerchuk
Prof. CS dept. Central Washington University
http://www.cwu.edu/~borisk

>>> On 11/1/2010 at 5:04 PM, in message <4C5F914ECF9F5F4CA2DB475171D055C3ACFD608837@ITDSRVMBX01.utep.edu>, "Kreinovich, Vladik" <vladik@utep.edu> wrote:

There are many ways to elicit a membership function. For example, you can ask an expert to estimate the degree to which a person is tall on a scale from 0 to 5. If the expert marks 3we take the value 3/5 = 0.6.

Alternatively, we can poll several experts, and if 6 out of 8 says that the guy is tall, we take 6/8 = 0.75.

There are many other ways all described in textbooks.

We are talking about fuzzy statements. For example, in fuzzy control we formalize statements of expert controllers who say "if the car in front is too close, brake a little bit". Similarly, in a medical system, we try to formalize the expert rule "if the tumor is much larger than 1 inch, order a surgery".

The proof is in using this system. The only alternative is NOT to use this knowledge at all. Fuzzy logic provides a way of using it, and often leads to good results. Practical application is the verification that you are looking for.

You use expert knowledge to design a fuzzy controller, then test it on a simulator, maybe tune it, and if it works well, use it in practice. Same thing with a fuzzy expert system: test it first on some past cases, tune if needed, and only after it has been validated, use it on real patients.

Hope it answers your concern.

-----Original Message-----
From: Jellat Arkadia [mailto:jellat007@hotmail.com]

It seems that Fuzzy Set theory does not use an empirically verifiable definition
of the membership function, in which sense, different people would be able to
apply different values to the same set of circumstances. In that sense, how could
we verify or falsify an statement like "This Man is tall". If this is used to judge how
serious one is sick, and would like to use the result to help doctors make decisions,
would this be problematic? As the membership function is not verified.

Alternatively, is it the case that this sort of membership function should only be
defined by the experts and used by the rest people?

Regards,
Nikola

*******************************************

>>> On 10/25/2010 at 4:48 PM, in message <4C5F914ECF9F5F4CA2DB475171D055C3ACFD608715@ITDSRVMBX01.utep.edu>, "Kreinovich, Vladik" <vladik@utep.edu> wrote:

Probability deals with precisely defined events. A coin either falls head or tail. If it falls tail in half of the cases, we say that the probability is 0.5.

It is not so easy to come up with a probability that the next person you meet is young, because the word is not precisely defined. Such natural language terms are used by experts all the time, so to describe them, we need fuzzy logic.

Fuzzy and probability describe two different aspects of uncertainty, in general, they can be (and should be) combined, since we often have both the knowledge about the probabilities and knowledge described in imprecise ("fuzzy") terms of natural language.

There is fuzzy reasoning, but it is NOT a substitute for probabilistic reasoning -- probabilistic reasoning starts with probabilities, fuzzy reasoning starts with imprecise fuzzy terms.

This is a basic description, but of course, you need to read some papers and textbooks to get into technical details.

****************************************************************************************************************

Dear BISC members,

I'm an undergraduate student from the UK and I'm very interested
in fuzzy set theory. As a beginner, I would like to ask the following
questions:

Q1. What is the difference between probability and fuzzy set?

Q2. Can fuzzy set theory be applied as a substitue of probabilistic
inference, to measure the likeliness something will happen or
be classified into category X?

Regards,
Nikola

Question from an undergraduate student

2010-11-06T01:05:00.001-07:00

From: Boris Kovalerchuk <Boris.Kovalerchuk@cwu.edu>
Date: 2010/11/5
Subject: [bisc-group] RE: Question from an undergraduate student
To: "bisc-group@lists.eecs.berkeley.edu" <bisc-group@lists.eecs.berkeley.edu>
Cc: "jellat007@hotmail.com" <jellat007@hotmail.com>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear BISC members,

Let me play a devil's advocate relative to the undergraduate student's questions on relations between fuzzy sets and probability. First I would rather agree with Hans that answers on differences in the "zoo" of uncertainties could be better.

There is a popular idea that "Old" and "Young" are imprecise overlapping concepts and therefore need to be modeled by using fuzzy set membership functions. ("Such natural language terms are used by experts all the time, so to describe them, we need fuzzy logic." V. Kreinovich, BISC, 10/25/2010).

It is commonly contrasted with the probability theory, which deals with crisp disjoint (mutually exclusive) elementary events, e.g., die sides. This was stated as a major difference between the fuzzy set theory and the probability theory. "Logic and probability assume "crisp" predicates and classical set-theoretic semantics" (R. Loui, BISC, 11/2/2010).

This answer is incomplete and leads us to the conceptual difficulties. As often the case, the devil is in the details. This answer did not tell us how to construct these Membership Functions (MFs).

Nikola asked that question recently. Vladik Kreinovich answered that one of the ways is using frequencies of subjective human answers (".we can poll several experts, and if 6 out of 8 says that the guy is tall, we take 6/8 = 0.75."). The fuzzy literature is full of such frequencies for computing MFs for ages Young, Old, Middle Age etc, e.g., Hall, Kandel, Szabo,1986.

If we construct MFs by using such frequencies then it seems that we start to contradict to ourselves.

We denied probabilistic approach for modeling overlapping imprecise linguistic terms and then use it via frequencies to build fuzzy membership functions for these terms.

Now about the next devil in this zoo. If we produce MFs in a probabilistic way (frequency-based or subjective), why do we not use the probabilistic operations to combine them (union and intersection) but use multiple T-norms and T-conorms?

Next, we have at least three flavors of probability: (1) frequency-based, (2) subjective, and (2) axiomatic (Kolmogorov's axioms). The last one abstracts the first two interpretations and maybe others yet unknown. Thus, thinking about probabilities only as frequencies of repeating events is a very narrow view of probability that should be avoided when we try to distinguish fuzzy membership functions from probabilities.

Comments welcome.

Regards

Boris

Boris Kovalerchuk

Prof. CS dept. Central Washington University

http://www.cwu.edu/~borisk

>>> On 11/1/2010 at 5:04 PM, in message <4C5F914ECF9F5F4CA2DB475171D055C3ACFD608837@ITDSRVMBX01.utep.edu>, "Kreinovich, Vladik" <vladik@utep.edu> wrote:

There are many ways to elicit a membership function. For example, you can ask an expert to estimate the degree to which a person is tall on a scale from 0 to 5. If the expert marks 3we take the value 3/5 = 0.6.

Alternatively, we can poll several experts, and if 6 out of 8 says that the guy is tall, we take 6/8 = 0.75.

There are many other ways all described in textbooks.

We are talking about fuzzy statements. For example, in fuzzy control we formalize statements of expert controllers who say "if the car in front is too close, brake a little bit". Similarly, in a medical system, we try to formalize the expert rule "if the tumor is much larger than 1 inch, order a surgery".

The proof is in using this system. The only alternative is NOT to use this knowledge at all. Fuzzy logic provides a way of using it, and often leads to good results. Practical application is the verification that you are looking for.

You use expert knowledge to design a fuzzy controller, then test it on a simulator, maybe tune it, and if it works well, use it in practice. Same thing with a fuzzy expert system: test it first on some past cases, tune if needed, and only after it has been validated, use it on real patients.

Hope it answers your concern.

-----Original Message-----
From: Jellat Arkadia [mailto:jellat007@hotmail.com]

It seems that Fuzzy Set theory does not use an empirically verifiable definition
of the membership function, in which sense, different people would be able to
apply different values to the same set of circumstances. In that sense, how could
we verify or falsify an statement like "This Man is tall". If this is used to judge how
serious one is sick, and would like to use the result to help doctors make decisions,
would this be problematic? As the membership function is not verified.

Alternatively, is it the case that this sort of membership function should only be
defined by the experts and used by the rest people?

Regards,
Nikola

*******************************************

>>> On 10/25/2010 at 4:48 PM, in message <4C5F914ECF9F5F4CA2DB475171D055C3ACFD608715@ITDSRVMBX01.utep.edu>, "Kreinovich, Vladik" <vladik@utep.edu> wrote:

Probability deals with precisely defined events. A coin either falls head or tail. If it falls tail in half of the cases, we say that the probability is 0.5.

It is not so easy to come up with a probability that the next person you meet is young, because the word is not precisely defined. Such natural language terms are used by experts all the time, so to describe them, we need fuzzy logic.

Fuzzy and probability describe two different aspects of uncertainty, in general, they can be (and should be) combined, since we often have both the knowledge about the probabilities and knowledge described in imprecise ("fuzzy") terms of natural language.

There is fuzzy reasoning, but it is NOT a substitute for probabilistic reasoning -- probabilistic reasoning starts with probabilities, fuzzy reasoning starts with imprecise fuzzy terms.

This is a basic description, but of course, you need to read some papers and textbooks to get into technical details.

****************************************************************************************************************

Dear BISC members,

I'm an undergraduate student from the UK and I'm very interested
in fuzzy set theory. As a beginner, I would like to ask the following
questions:

Q1. What is the difference between probability and fuzzy set?

Q2. Can fuzzy set theory be applied as a substitue of probabilistic
inference, to measure the likeliness something will happen or
be classified into category X?

Regards,
Nikola

Kreinovich, Vladik

2010-11-06T01:03:00.001-07:00

From: Kreinovich, Vladik <vladik@utep.edu>
Date: 6 November 2010 00:48
Subject: [bisc-group]
To: "bisc-group@lists.eecs.berkeley.edu" <bisc-group@lists.eecs.berkeley.edu>
Cc: "Boris.Kovalerchuk@cwu.EDU" <Boris.Kovalerchuk@cwu.edu>, "hunguyen@nmsu.edu" <hunguyen@nmsu.edu>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Boris is absolutely right, this is where a random set interpretation of fuzzy sets comes in -- the one advocated by Hung T. Nguyen and many others.

-----Original Message-----
From: Boris Kovalerchuk [mailto:Boris.Kovalerchuk@cwu.EDU]
Sent: Thursday, November 04, 2010 11:19 PM
To: bisc-group@lists.eecs.berkeley.edu
Cc: 'jellat007@hotmail.com'

Dear BISC members,

Let me play a devil*s advocate relative to the undergraduate student*s
questions on relations between fuzzy sets and probability. First I would
rather agree with Hans that answers on differences in the *zoo* of
uncertainties could be better.

There is a popular idea that *Old* and *Young* are imprecise
overlapping concepts and therefore need to be modeled by using fuzzy set
membership functions. (*Such natural language terms are used by experts
all the time, so to describe them, we need fuzzy logic** V. Kreinovich,
BISC, 10/25/2010).

It is commonly contrasted with the probability theory, which deals with
crisp disjoint (mutually exclusive) elementary events, e.g., die sides.
This was stated as a major difference between the fuzzy set theory and
the probability theory. *Logic and probability assume "crisp" predicates
and classical set-theoretic semantics* (R. Loui, BISC, 11/2/2010).

This answer is incomplete and leads us to the conceptual difficulties.
As often the case, the devil is in the details. This answer did not
tell us how to construct these Membership Functions (MFs).
Nikola asked that question recently. Vladik Kreinovich answered that
one of the ways is using frequencies of subjective human answers (**we
can poll several experts, and if 6 out of 8 says that the guy is tall,
we take 6/8 = 0.75.*). The fuzzy literature is full of such frequencies
for computing MFs for ages Young, Old, Middle Age etc, e.g., Hall,
Kandel, Szabo,1986.

If we construct MFs by using such frequencies then it seems that we
start to contradict to ourselves.
We denied probabilistic approach for modeling overlapping imprecise
linguistic terms and then use it via frequencies to build fuzzy
membership functions for these terms.

Now about the next devil in this zoo. If we produce MFs in a
probabilistic way (frequency-based or subjective), why do we not use the
probabilistic operations to combine them (union and intersection) but
use multiple T-norms and T-conorms?

Next, we have at least three flavors of probability: (1)
frequency-based, (2) subjective, and (2) axiomatic (Kolmogorov*s
axioms). The last one abstracts the first two interpretations and maybe
others yet unknown. Thus, thinking about probabilities only as
frequencies of repeating events is a very narrow view of probability
that should be avoided when we try to distinguish fuzzy membership
functions from probabilities.

Comments welcome.

Regards

Boris

Boris Kovalerchuk
Prof. CS dept. Central Washington University
http://www.cwu.edu/~borisk

>>> On 11/1/2010 at 5:04 PM, in message
<4C5F914ECF9F5F4CA2DB475171D055C3ACFD608837@ITDSRVMBX01.utep.edu>,
"Kreinovich, Vladik" <vladik@utep.edu> wrote:
There are many ways to elicit a membership function. For example, you
can ask an expert to estimate the degree to which a person is tall on a
scale from 0 to 5. If the expert marks 3we take the value 3/5 = 0.6.

Alternatively, we can poll several experts, and if 6 out of 8 says that
the guy is tall, we take 6/8 = 0.75.

There are many other ways all described in textbooks.

We are talking about fuzzy statements. For example, in fuzzy control we
formalize statements of expert controllers who say "if the car in front
is too close, brake a little bit". Similarly, in a medical system, we
try to formalize the expert rule "if the tumor is much larger than 1
inch, order a surgery".

The proof is in using this system. The only alternative is NOT to use
this knowledge at all. Fuzzy logic provides a way of using it, and often
leads to good results. Practical application is the verification that
you are looking for.

You use expert knowledge to design a fuzzy controller, then test it on
a simulator, maybe tune it, and if it works well, use it in practice.
Same thing with a fuzzy expert system: test it first on some past cases,
tune if needed, and only after it has been validated, use it on real
patients.

Hope it answers your concern.

-----Original Message-----
From: Jellat Arkadia [mailto:jellat007@hotmail.com]

It seems that Fuzzy Set theory does not use an empirically verifiable
definition
of the membership function, in which sense, different people would be
able to
apply different values to the same set of circumstances. In that sense,
how could
we verify or falsify an statement like "This Man is tall". If this is
used to judge how
serious one is sick, and would like to use the result to help doctors
make decisions,
would this be problematic? As the membership function is not verified.

Alternatively, is it the case that this sort of membership function
should only be
defined by the experts and used by the rest people?

Regards,
Nikola

*******************************************

>>> On 10/25/2010 at 4:48 PM, in message
<4C5F914ECF9F5F4CA2DB475171D055C3ACFD608715@ITDSRVMBX01.utep.edu>,
"Kreinovich, Vladik" <vladik@utep.edu> wrote:
Probability deals with precisely defined events. A coin either falls
head or tail. If it falls tail in half of the cases, we say that the
probability is 0.5.

It is not so easy to come up with a probability that the next person
you meet is young, because the word is not precisely defined. Such
natural language terms are used by experts all the time, so to describe
them, we need fuzzy logic.

Fuzzy and probability describe two different aspects of uncertainty, in
general, they can be (and should be) combined, since we often have both
the knowledge about the probabilities and knowledge described in
imprecise ("fuzzy") terms of natural language.

There is fuzzy reasoning, but it is NOT a substitute for probabilistic
reasoning -- probabilistic reasoning starts with probabilities, fuzzy
reasoning starts with imprecise fuzzy terms.

This is a basic description, but of course, you need to read some
papers and textbooks to get into technical details.

****************************************************************************************************************

Dear BISC members,

I'm an undergraduate student from the UK and I'm very interested
in fuzzy set theory. As a beginner, I would like to ask the following
questions:

Q1. What is the difference between probability and fuzzy set?

Q2. Can fuzzy set theory be applied as a substitue of probabilistic
inference, to measure the likeliness something will happen or
be classified into category X?

Regards,
Nikola

Dear Members of the BISC Group:

2010-11-02T00:56:00.001-07:00

Dear Members of the BISC Group:

        Sometime ago I made a prediction: Eventually, the hidden-from-view conflict between Putin and Medvedev will come into the open. At this point, my expectation is that the conflict will come into the open in the near future. More concretely, the conflict is likely to be precipitated by Medvedev's pardoning Khodarkowsky, the imprisoned Russian Oligarch who was a bitter enemy of Putin. Watch the headlines. Comments are welcome.

                        Regards,

                        Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

Apostolos Syropoulos --- Question from an undergraduate student

2010-11-02T00:54:00.001-07:00

---------- Forwarded message ----------
From: Jellat Arkadia <jellat007@hotmail.com>
Date: 2010/11/1

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear BISC members,

Here comes my question again.

It seems that Fuzzy Set theory does not use an empirically verifiable definition
of the membership function, in which sense, different people would be able to
apply different values to the same set of circumstances. In that sense, how could
we verify or falsify an statement like "This Man is tall". If this is used to judge how
serious one is sick, and would like to use the result to help doctors make decisions,
would this be problematic? As the membership function is not verified.

Alternatively, is it the case that this sort of membership function should only be
defined by the experts and used by the rest people?

Regards,
Nikola

Date: Thu, 28 Oct 2010 11:18:28 +0300
From: ijdt.editor@gmail.com
To: amgsolo@mavericktechnologies.us
CC: bisc-group@lists.eecs.berkeley.edu; jellat007@hotmail.com
Subject: Re: [bisc-group] Question from an undergra! duate student

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

2010/10/27 Ashu M. G. Solo > ********************************************************************* > Berkeley Initiative in Soft Computing (BISC) > ********************************************************************* > > To answer Nikola's question, probability can be used for type one > uncertainty, as described below, whereas fuzzy logic can be used for type > two uncertainty, as described below. > Hello, Recently, I read the papers on the so-called Free Will Theorem, see http://en.wikipedia.org/wiki/Free_will_theorem for links to PDFs of the papers. Roughly, the authors argue that randomness is a derived property and not a fundamental one. Indeed, they say that they agree with Einstein's famous credo that God does not play dices. But they do not deny fuzziness, which IMHO is something very important. A.S. -- Apostolos Syropoulos Xanthi, GREECE

2010/10/27 Ashu M. G. Solo <amgsolo@mavericktechnologies.us>

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

To answer Nikola's question, probability can be used for type one
uncertainty, as described below, whereas fuzzy logic can be used for type
two uncertainty, as described below.

Hello,

Recently, I read the papers on the so-called Free Will Theorem, see

http://en.wikipedia.org/wiki/Free_will_theorem

for links to PDFs of the papers. Roughly, the authors argue that randomness
is a derived property and not a fundamental one. Indeed, they say that they agree
with Einstein's famous credo that God does not play dices. But they do not deny
fuzziness, which IMHO is something very important.

A.S.

--
Apostolos Syropoulos
Xanthi, GREECE

The explanation that follows is an excerpt from the following book chapter:

2010-10-27T00:26:00.001-07:00

The explanation that follows is an excerpt from the following book chapter:

Gupta, Madan M., and Solo, Ashu M. G. [2007], "Uncertainty in Computational
Perception and Cognition," Forging New Frontiers: Fuzzy Pioneers I:
Studies in Fuzziness and Soft Computing, Nikravesh, M., Kacprzyk, J., and
Zadeh, L. A., editors, Springer Verlag, pp. 251-266.

There are various types of uncertainty. However, they can be classified
under two broad categories [1]: type one uncertainty and type two
uncertainty.

3.1 Type One Uncertainty

Type one uncertainty deals with information that arises from the random
behavior of physical systems. The pervasiveness of this type of uncertainty
can be witnessed in random vibrations of a machine, random fluctuations of
electrons in a magnetic field, diffusion of gases in a thermal field, random
electrical activities of cardiac muscles, uncertain fluctuations in the
weather pattern, and turbulent blood flow through a damaged cardiac valve.
Type one uncertainty has been studied for centuries. Complex statistical
mathematics has evolved for the characterization and analysis of such random
phenomena.

3.2 Type Two Uncertainty

Type two uncertainty deals with information or phenomena that arise from
human perception and cognitive processes or from cognitive information in
general. This subject has received relatively little attention. Perception
and cognition through biological sensors (eyes, ears, nose, etc.),
perception of pain, and other similar biological events throughout our
nervous system and neural networks deserve special attention. The
perception and cognition phenomena associated with these processes are
characterized by many great uncertainties and cannot be described by
conventional statistical theory. A person can linguistically express
perceptions experienced through the senses, but these perceptions cannot be
described using conventional statistical theory.

-----Original Message-----
From: Kreinovich, Vladik [mailto:vladik@utep.edu]
Sent: Monday, October 25, 2010 5:49 PM
To: bisc-group@lists.eecs.berkeley.edu; 'jellat007@hotmail.com'
Subject: RE: [bisc-group] Question from an undergraduate student

Probability deals with precisely defined events. A coin either falls head or
tail. If it falls tail in half of the cases, we say that the probability is
0.5.

It is not so easy to come up with a probability that the next person you
meet is young, because the word is not precisely defined. Such natural
language terms are used by experts all the time, so to describe them, we
need fuzzy logic.

Fuzzy and probability describe two different aspects of uncertainty, in
general, they can be (and should be) combined, since we often have both the
knowledge about the probabilities and knowledge described in imprecise
("fuzzy") terms of natural language.

There is fuzzy reasoning, but it is NOT a substitute for probabilistic
reasoning -- probabilistic reasoning starts with probabilities, fuzzy
reasoning starts with imprecise fuzzy terms.

This is a basic description, but of course, you need to read some papers and
textbooks to get into technical details.

****************************************************************************
************************************

Dear BISC members,

I'm an undergraduate student from the UK and I'm very interested
in fuzzy set theory. As a beginner, I would like to ask the following
questions:

Q1. What is the difference between probability and fuzzy set?

Q2. Can fuzzy set theory be applied as a substitue of probabilistic
inference, to measure the likeliness something will happen or
be classified into category X?

Regards,
Nikola

Dear Members of the BISC Group:

2010-10-26T23:47:00.001-07:00

Dear Members of the BISC Group:

    In September 2010, the National Bureau of Economic Research (NBER) announced that the deepest recession since the Great Depression came to an end in June 2009. To say that the announcement was met with disbelief--particularly by those who were ravaged by the recession after June 2009--is an understatement. In defense of its announcement, NBER explained that they defined the end of the recession as the point at which recession reached its peak. If this is the case, then NBER should have announced that what had come to an end was not the recession but its growth. It is hard to understand how the peak of a recession can be defined as its end.

    Officially, recession is defined as a decline in GDP in two successive quarters. There are other definitions of recession, but all definitions are bivalent, that is, do not admit shades of truth. This is what I should like to criticize.

    Recession--like many other concepts in economics--is a fuzzy concept. What this implies is that recession is a matter of degree. A realistic definition of recession should be based on fuzzy logic--not on bivalent logic. Basically, what is needed is a definition in the spirit of the Richter scale. Bivalent definitions, no matter what they are, lead to counterintuitive conclusions. The NBER announcement is an example.

    In a 1976 paper entitled "A fuzzy-algorithmic approach to the definition of complex or imprecise concepts." I drew attention to the deep-seated tradition in science of defining complex concepts as bivalent concepts in closed form. The problem is that, more often than not, such definitions are unrealistic. I argued that realistic definitions must be algorithmic and based on fuzzy logic. This applies, in particular, to the concept of a recession.

          Regards to all.

          Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

Probability deals with precisely defined events

2010-10-26T00:12:00.001-07:00

	Kreinovich, Vladik <vladik@utep.edu>

Probability deals with precisely defined events. A coin either falls head or tail. If it falls tail in half of the cases, we say that the probability is 0.5.

It is not so easy to come up with a probability that the next person you meet is young, because the word is not precisely defined. Such natural language terms are used by experts all the time, so to describe them, we need fuzzy logic.

Fuzzy and probability describe two different aspects of uncertainty, in general, they can be (and should be) combined, since we often have both the knowledge about the probabilities and knowledge described in imprecise ("fuzzy") terms of natural language.

There is fuzzy reasoning, but it is NOT a substitute for probabilistic reasoning -- probabilistic reasoning starts with probabilities, fuzzy reasoning starts with imprecise fuzzy terms.

This is a basic description, but of course, you need to read some papers and textbooks to get into technical details.

****************************************************************************************************************

Dear BISC members,

I'm an undergraduate student from the UK and I'm very interested
in fuzzy set theory. As a beginner, I would like to ask the following
questions:

Q1. What is the difference between probability and fuzzy set?

Q2. Can fuzzy set theory be applied as a substitue of probabilistic
inference, to measure the likeliness something will happen or
be classified into category X?

Regards,
Nikola

Invention vs discovery

2010-10-23T01:23:00.001-07:00

Dear Marianne and Vesa,

I am very interested in your discussion on creation and invention. I
believe Lotfi's insights on soft computing, fuzzy logic, as well as
computational intelligence have provided us for fundamental mathematical
preparations for possible computational simulations of this fantastic
phenomenon of the human brain.

The most classic book on this topic was written by W. I. Beveridge (1957)
entitled "The Art of Scientific Investigation."

In the field of cognitive informatics and cognitive computing, I tried to
investigate into creativity in the following paper:

Wang, Y. (2009), "On Cognitive Foundations of Creativity and the Cognitive
Process of Creation", International Journal of Cognitive Informatics and
Natural Intelligence, 3(4), 1-18. (See:
http://enel.ucalgary.ca/People/wangyx/Publications/Papers/CI/IJCINI-3401-CogCreativity.pdf
or
http://enel.ucalgary.ca/People/wangyx/Research/Research.htm .)

"In the article, I perceived that creativity is a gifted ability of human
beings in thinking, inference, problem solving, and product development. A
creation is a new and unusual relation between two or more objects that
generates a novel and meaningful concept, solution, method, explanation,
or product. This article formally investigates into the cognitive process
of creation and creativity as one of the most fantastic life functions.
The cognitive foundations of creativity are explored in order to explain
the space of creativity, the approaches to creativity, the relationship
between creation and problem solving, and the common attributes of
inventors. A set of mathematical models of creation and creativity is
established on the basis of the tree structures and properties of human
knowledge known as concept trees. The measurement of creativity is
quantitatively analyzed, followed by the formal elaboration of the
cognitive process of creation as a part of the Layered Reference Model of
the Brain (LRMB) in this article.

Best regards,

Yingxu Wang

> Dr. Vesa A. Niskanen wrote:
>
> Dear Marianne,
>
> You clarified well the distinction. At the core of these phenomena is
> the human insight. In science this subject matter leads to the problems
> concerning the hypotheses, i.e., how do we establish our hypotheses.
> Quite often random events, good luck, imagination (and dreams) and
> intuition play essential role in this context. This process is sometimes
> referred to as serendipity.
>
> As we know, such "mechanical" reasoning methods as induction, abduction
> (retroduction) and reasoning by analogy may also help us to generate new
> hypotheses on some occasions.
>
> Vesa
>
>
> *************************************************
> Sincerely,
>
> Mr. Vesa A. Niskanen
> Ph. D. (Assoc. Prof. / Docent)
>
> University of Helsinki
> Dept. of Economics & Management (Taloustieteen laitos)
> PO Box 27, 00014 Helsinki, Finland
> (Street address: Viikki Campus, Latokartanonkaari 9,
> Building A, room 223)
> Tel. +358 50 415 1204 (mobile/office hours), +358 40 503 2031
> (mobile/private & business)
> E-mail: vesa.a.niskanen [ät] helsinki.fi
> Website: www.mm.helsinki.fi/users/niskanen
> SKYPE: vaniskanen
>
> "Omnia mea mecum porto"
> *************************************************
>
> BELIS Marianne wrote:
>> Inventing amounts to combine existing entities in a new structure with
>> new properties. Examples:
>> - the inventor of the wheel assembled rolling objects (probably tree
>> trunks) in a new way for transporting objects.
>> - Zadeh combined existing concepts (bivalent sets, membership, gradual
>> changes) in a new entity (fuzzy sets) with new properties.
>> Discovering amounts to unveil structures already existing with their
>> properties. Examples:
>> - Herschel discovered the planet Uranus.
>> - Fleming discovered the penicillin (the antibacterial properties of the
>> mold spores)
>> Inventions and discoveries mutually influence each other: the discovery
>> of existing structures leads to a greater variety of new combinations
>> and the invention of new structures enlarge the possibilities of
>> exploring existing objects.
>>
>> Marianne Belis
>>
>>
>

This is a philosophical comment to heuristics

2010-10-14T23:52:00.000-07:00

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

This is a philosophical comment to heuristics:

- The term itself has Greek origin (to invent).

- Heuristics comprises at least two constituents: inventing and discovering. Eg. Prof. Zadeh invented fuzzy sets, whereas Herschel discovered the planet Uranus.

- rediscovering is also possible

- Heuristics has been studied in psychology within creative thinking. This area still still having many black holes. We can also study the logic of invention/discovery but this has been an even more challenging problem.

Fuzzy feelings,

*************************************************
Sincerely,

Mr. Vesa A. Niskanen
Ph. D. (Assoc. Prof. / Docent)

University of Helsinki
Dept. of Economics & Management (Taloustieteen laitos)
PO Box 27, 00014 Helsinki, Finland
(Street address: Viikki Campus, Latokartanonkaari 9,
Building A, room 223)
Tel. +358 50 415 1204 (mobile/office hours), +358 40 503 2031
(mobile/private & business)
E-mail: vesa.a.niskanen [ät] helsinki.fi
Website: www.mm.helsinki.fi/users/niskanen
SKYPE: vaniskanen

"Omnia mea mecum porto"
*************************************************

Perlovsky, Leonid Civ USAF AFMC AFRL/RYHE wrote:

Dear Stuart,

This are all interesting topic, and I would be glad to continue. But first, I made no comparison between heuristics and Boolean logic.
I'll be glad to hear your definition of heuristics.

Wiki defines them as "experience-based techniques for problem solving."

I used the word the way it is widely used currently after Tversky-Kahneman research, Wiki:

" Although much of the work of discovering heuristics in human decision-makers was done by Amos Tversky and Daniel Kahneman[3], the concept has been originally introduced by Nobel laureate Herbert Simon. Gerd Gigerenzer focuses on how heuristics can be used to make judgments that are in principle accurate, rather than producing cognitive biases – heuristics that are "fast and frugal".[4]

This emphasizes understanding of heuristics as ready-made rules, which have not been adapted to the current situation.

If you have any idea of where new knowledge is coming from, please let us know. Because most of people cannot answer this question scientifically.

When you say SYMBOLIC, do you mean a mathematical notation with axiomatically given meaning, or a psychological process connecting conscious and unconscious? - if the later one, please give mathematical definition (or at least an approach).

I am looking forward to an interesting discussion.

Best
Leonid

-----Original Message-----
From: Rubin, Stuart H CIV SPAWARSYSCEN-PACIFIC, 56250 [mailto:stuart.rubin@navy.mil] Sent: Thursday, October 07, 2010 7:46 PM
To: bisc-group@lists.eecs.berkeley.edu
Subject: RE: [bisc-group] On problems and solutions - not necessarily in this order
Excuse me Leonid, but heuristics are a symbolic mathematical construct.
Their use results in inadmissible search. While it is known how they fit any existing data set, it is not known how they will interrelate otherwise. They are not a logic and thus are not properly compared to Boolean logic. The second-order predicate calculus is incomplete because it is capable of self-reference. No such analysis can be made for heuristics. They cannot be included in Learnability Theory because they don't analyze as do formal logics.

v/r,
Stuart Rubin, Ph.D.
SSC-PAC
Tel. (619) 553-3554
Fx. (619) 553-1130
E. stuart.rubin@navy.mil

The problems you work on are very difficult to be sure; but, it is even more difficult to know what problems to work on.
---Stuart---

-----Original Message-----
From: Perlovsky, Leonid Civ USAF AFMC AFRL/RYHE [mailto:Leonid.Perlovsky@hanscom.af.mil]
Sent: Thursday, October 07, 2010 10:08 AM
To: bisc-group@lists.eecs.berkeley.edu
Subject: RE: [bisc-group] On problems and solutions - not necessarily in this order
Stuart, Michelle and everybody,

I would like to emphasize that heuristics might be useful, but they rely on what is known. Same Boolean logic. These techniques do not create new knowledge.

Fuzzy logic can create new knowledge, like the mind sometimes does.

Best
Leonid

-----Original Message-----
From: Rubin, Stuart H CIV SPAWARSYSCEN-PACIFIC, 56250 [mailto:stuart.rubin@navy.mil]
Sent: Tuesday, October 05, 2010 9:28 PM
To: bisc-group@lists.eecs.berkeley.edu
Subject: RE: [bisc-group] On problems and solutions - not necessarily in this order
Michelle:

They say that hindsight is 20/20. One cannot know for sure that one has selected the right problem to work on or that one has selected the right time to work on it. Theoretically, this is an unsolvable problem. Its pragmatic solution entails the use of heuristics and herein lies an interesting version of self reference; namely, heuristics is to exhaustive and otherwise intractable search what fuzzy logic is to Boolean logic; again with my humble complements to Lotfi.
v/r,
Stuart Rubin, Ph.D.
SSC-PAC
Tel. (619) 553-3554
Fx. (619) 553-1130
E. stuart.rubin@navy.mil

The problems you work on are very difficult to be sure; but, it is even more difficult to know what problems to work on.
---Stuart---

-----Original Message-----
From: Michelle Quirk [mailto:pal@math.utexas.edu]
Sent: Tuesday, October 05, 2010 4:13 PM
To: bisc-group@lists.eecs.berkeley.edu
Subject: [bisc-group] On problems and solutions - not necessarily in this order

Stuart,

Not to worry: once one has a solution, anything is simple - and one certainly does have a solution when one has difficulties selecting the
problem(s) to work on.

All in all your note makes me of course think of the Library of Babel by
Borges: there must be somewhere the book of books - the captured wisdom of all wisdoms - and the solution to all problems - even those problems yet to be formulated.

I would like to close with the due tribute to Russell (for his "the set of all sets" including fuzzy ones) and to Turing: we would not be writing today all these elevated thoughts - we might have been delayed by a few moments.

Best,
Michelle

On Tue, 5 Oct 2010, Rubin, Stuart H CIV SPAWARSYSCEN-PACIFIC, 56250
wrote:

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

--
Michelle Quirk

LOTFI ZADEH: Factual Information about the Impact of Fuzzy Logic

2010-10-04T23:33:00.001-07:00

Factual Information about the Impact of Fuzzy Logic

L.A. ZADEH

PATENTS

Number of fuzzy-logic-related patents applied for in Japan: 17,740
Number of fuzzy-logic-related patents issued in Japan: 4,801
Number of fuzzy-logic-related patents issued in the US: around 1,700

JOURNALS

Fuzzy in title

Fuzzy Sets and Systems
IEEE Transactions on Fuzzy Systems
Fuzzy Optimization and Decision Making

4. Journal of Intelligent & Fuzzy Systems

Fuzzy Economic Review
International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems
Journal of Japan Society for Fuzzy Theory and Systems
International Journal of Fuzzy Systems
International Review of Fuzzy Mathematics
Fuzzy Systems and Soft Computing
Turkish Journal of Fuzzy Systems
Annals of Fuzzy Sets, Fuzzy Logic and Fuzzy Systems

COUNT OF PUBLICATIONS

Count of publications containing the word "fuzzy" in the title, as cited in INSPEC and MATH.SCI.NET databases.

Compiled on October 4, 2010.

INSPEC Database

1970-1979: 521

1980-1989: 2,163

1990-1999: 20,210

2000-present: 48,627

Total: 71,521

MathSciNet Database

1970-1979: 444

1980-1989: 2,466

1990-1999: 5,487

2000-present: 10,439

Total: 18,863

Number of citations of papers by L.A. Zadeh (Web of Science Citation): 28,122

2010 incomplete

Number of citations of papers by L.A. Zadeh (Google Scholar): 85,500

Warm regards.

Lotfi

 --  Lotfi A. Zadeh

Professor in the Graduate School Director,

Berkeley Initiative in Soft Computing (BISC)

Address:  729 Soda Hall #1776

Computer Science Division

Department of Electrical Engineering and Computer Sciences

University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

Dear Prof. Dr.Zimmermann

2010-10-01T03:04:00.001-07:00

Dear Prof. Dr.Zimmermann

I have working on Fraud/Crime/ Virtual Customer detection techniques
based on Data Mining since 2006. I proposed a new algorithm for fraud
Detection in Streaming Data (ISI journal) and implemented a new system
for that.
Please check my profile and full information and my CV in the :
http://sites.google.com/site/frauddetectionandstreammining/

I like continue my study in this field. Please guide me.

Thanks in advance
Best Regards
Naeimeh Laleh

Dear Members of the BISC Group, Video clip

2010-10-01T00:22:00.001-07:00

Dear Members of the BISC Group,

    Hamid Berenji brought to my attention a video clip related to my receiving the Benjamin Franklin Institute Medal in 2009. I saw it for the first time today. http://www.youtube.com/watch?v=2ScTwFCcXGo&NR=1&feature=fvwp

    Regards to all.

    Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

From: Lotfi A. Zadeh zadeh@eecs.berkeley.edu

2010-09-28T00:41:00.001-07:00

Dear Members of the BISC Group:

        There is an interesting application of Z-mouse which involves construction of the convex hall of a fuzzy set. If you know of articles in which this problem is addressed, I would appreciate it if you could bring them to my attention.

      Regards to all

      Lotfi

--  Lotfi A. Zadeh  Professor in the Graduate School Director, Berkeley Initiative in Soft Computing (BISC)   Address:  729 Soda Hall #1776 Computer Science Division Department of Electrical Engineering and Computer Sciences University of California  Berkeley, CA 94720-1776  zadeh@eecs.berkeley.edu  Tel.(office): (510) 642-4959  Fax (office): (510) 642-1712  Tel.(home): (510) 526-2569  Fax (home): (510) 526-2433  URL: http://www.cs.berkeley.edu/~zadeh/  BISC Homepage URLs URL: http://zadeh.cs.berkeley.edu/

From: yoshiki uemura

2010-09-27T23:15:00.001-07:00

From: yoshiki uemura <uemura0742@yahoo.co.jp>
Date: 2010/9/27

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear Prof. Dr. Zimmermann

Would you remmenber me in Archen? And with you, I agreed with the Banket.

I was the potential refree for Fuzzy Sets and Systems.

My note would apear in Fuzzy Sets and Systems?

Thank you for the very happy night with the Archen.

First of all, I finished my email.

King Regards

Yoshiki Uemura

Hans-Juergen Zimmermann <Hans-Juergen.Zimmermann@inform-ac.com> wrote:

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************
Dear BISC Members:
Lotfi Zadeh remarked recently, that the members of BISC might be
interested in real applications of Soft Computing and Operations
Research. In 1969 I founded a company, Institute for Operations Research
and Management, Inform GmbH, which focuses only on real applications of
Operations Research ,and later Soft Computing, in order to prove its
applicability.In the meantime this company employs appr. 400 people and
has saved industry appr. 15 billions 竄ャ. Whoever is interested in this
could look inti its webside :www.inform-ac.com. Have fun and
Best regards,
Hans (Zimmermann)

Prof.em.Dr.Dr.h.c.mult.H.-J.Zimmermann
INFORM GmbH
Pascalstr.23
D-52076 Aachen (Germany)
URL:http://www.or.rwth-aachen.de/Zimmermann/ (
http://www.or.rwth-aachen.de/Zimmermann/ )

B. Schweizer obituary

2010-09-25T02:35:00.001-07:00

From: Itziar García Honrado <itziar.garcia@softcomputing.es>

Berthold Schweizer (1929-2010)

Berthold Schweizer was born in Germany in a Jewish family that emigrated to United States of America in 1937. He became interested in Mathematics as a child, and studied in the Massachusetts Institute of Technology (B.S. 1951) and at t e Illinois Institute of Technology in Chicago (M.S. 1954). At this institution he earned his Ph.D. in 1956 under Karl Menger. Berthold Schweizer had a long an successful academic career. He taught at the Naval Ordnance laboratory (1951), IIT (1956-57), San Diego State College (1957-58), UCLA (1958-60), the University of Arizona (1960-65 and 1968-70) and the University of Massachusetts at Amherst (1965-68) and from 1970 until his retirement in 1996. He had several visiting appointments in Vienna, Perugia, Barcelona, Lecce, Bern and Beijing. He was a superb expositor and a motivating teacher.

He was the adviser of 12 PhD students and also mentored many post-doctoral students and visitors from the United States and various foreign countries.

Berthold Schweizer and Abe Sklar were the founders of the theory of probabilistic metric spaces, a concept introduced by Karl Menger in 1942 that Schweizer, Sklar, and their students developed. It was in this mathematical elaboration of the theory of probabilistic metric spaces that Schweizer and Sklar developed a remarkable body of new mathematical machinery, using techniques from the theory of functional equations and inequalities, and making important contributions to the study of the associative functions in real intervals (t-norms, t-conorms) and the theory of copulas (created by Abe Sklar).

Berthold Schweizer wrote many papers and some fundamental books, working always with Abe Sklar or his students. In the last years he was working with the Abe Sklar on Menger's legacy and a new project concerning what they called the grammar of functions.

It is interesting to note that beyond the interest of probabilistic metric spaces, Schweizer's work had a profound impact in other fields: functional equations, inequalities, fuzzy sets theory, mathematical models in economics, etc.

Berthold Schweizer was a great mathematician, a loving husband, father and grandfather, and a very warm, open, and generous man. I had the opportunity to work with him in 1976-77 and since then I have had the pleasure to do scientific work with him for more than 32 years and to develop a close personal relationship. He was an inspiring adviser and a wonderful person. We will miss him very much, but our good memories and his mathematical creativity will remain with us for years to come.

                                                             Claudi Alsina
                                                           Universitat Politècnica de Catalunya
                                                             claudio.alsina@upc.edu

: [bisc-group] More on Z-mouse

2010-09-24T03:15:00.001-07:00

From: yoshiki uemura <uemura0742@yahoo.co.jp>
Date: 2010/9/23
Subject: Re: [bisc-group] More on Z-mouse

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************

Dear my members

Now, I proposed my short note "A Case Study of KIDO" for Cybernets and Systems.

As my true frends exsisted in this members, would you tell me whether my paper would appear in Fuzzy Sets and Systems or not?

Thank you

King Regards

Yoshiki Uemura

"Lotfi A. Zadeh" <zadeh@eecs.berkeley.edu> wrote:

*********************************************************************
Berkeley Initiative in Soft Computing (BISC)
*********************************************************************
Dear Members of the BISC Group:

    In a message posted to the BISC Group on July 30, 2010 I gave a brief description of the concept of a Z-mouse. The following is an elaboration of what was said in my earlier message. Comments are welcome.
The concept of a Z-mouse has deceptive simplicity. It takes a while to recognize that, in combination with the machinery of Computing with Words (CW or CWW), the concept of a Z-mouse carries the potential for a major impact on the ways in which uncertainty/imprecision is dealt with in science, engineering and other domains of human activity.

    Basically, a Z-mouse is a visual means for entry and retrieval of fuzzy data. A scale is shown on the screen of a computer. The cursor of a Z-mouse is a fuzzy circle, called an f-mark, which represents a fuzzy set. The cross section of an f-mark is a trapezoidal fuzzy set. The size of an f-mark is controlled by the user. To enter the value of a variable the user marks the scale with an f-mark centering on the scale. The size of an f-mark is a measure of the user's perception of uncertainty/imprecision about the value of the variable. A key point is that this uncertainty/imprecision is represented visually. The Z-mouse translates the f-mark internally into the membership function of a trapezoidal fuzzy set. This membership function serves as an object of computation in a CW engine. Computations are performed internally, unseen by the user. The result of computation is represented as an f-mark on a scale. The user can read, if needed, the parameters of the trapezoid.

    The importance of the Z-mouse derives from the fact that we live in a world of uncertainty/imprecision. A major source of uncertainty/imprecision is unsharpness of class boundaries, that is, fuzziness of class boundaries. In effect, fuzzy logic is the logic of classes with unsharp boundaries. In the world of uncertainty/imprecision, the Z-mouse serves as a means of visually representing one's perception of uncertainty/imprecision of the values of variables and degrees. As an illustration, suppose that I am asked: What is the probability that Obama will be reelected? I represent my perception of this probability as an f-mark. Suppose that I declare that I love oranges, and am asked to precisiate what I said. I do so by putting an f-mark on a scale. Suppose that I am asked if I am for or against legalization of drugs. I indicate the strength of my preference by using an f-mark. Suppose that I say: I believe that Robert is honest. To precisiate what I say I use an f-mark to represent my strength of believe and another f-mark to represent my perception of Robert's honesty. And so on.

    Essentially, the Z-mouse opens the door to: (a) a visual representation of uncertainty/imprecision for purposes of precisiation and communication; and (b) computation under uncertainty/imprecision through the use of the machinery of Computing with Words.

    A very important application area for the concept of a Z-mouse is decision analysis. Can you suggest other applications?

        Regards to all,

        Lotfi

P.S. A more detailed description of Z-mouse is included in my lecture "Precisiation of Meaning--Toward Computation with Natural Language," presented at IRI 2010 on August 4, 2010 in Las Vegas, Nevada. You can obtain a powerpoint file of my presentation from my website at http://www.cs.berkeley.edu/~zadeh/presentations.html
 --   Lotfi A. Zadeh   Professor in the Graduate School  Director, Berkeley Initiative in Soft Computing (BISC)      Address:   729 Soda Hall #1776  Computer Science Division  Department of Electrical Engineering and Computer Sciences  University of California   Berkeley, CA 94720-1776   zadeh@eecs.berkeley.edu   Tel.(office): (510)  642-4959   Fax (office): (510) 642-1712   Tel.(home): (510) 526-2569   Fax (home): (510) 526-2433   URL: http://www.cs.berkeley.edu/~zadeh/    BISC Homepage URLs  URL: http://zadeh.cs.berkeley.edu/  