Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
41_Toward a Generalized Theory of Uncertainty (GTU)--An Outline-2005

41_Toward a Generalized Theory of Uncertainty (GTU)--An Outline-2005

Ratings: (0)|Views: 14 |Likes:

More info:

Published by: Dr. Ir. R. Didin Kusdian, MT. on Mar 23, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

03/22/2010

pdf

text

original

 
Toward a generalized theoryof uncertainty (GTU)––an outline
Lotfi A. Zadeh
*
Berkeley initiative in Soft Computing (BISC), Computer Science Divisionand the Electronics Research Laboratory, Department of EECS, University of California,615 Soda Hall, Berkeley, CA 94720-1776, USA
Received 21 December 2004; accepted 26 January 2005Dedicated to Didier Dubois, Henri Prade and the memory of my mentors,Richard Bellman and Herbert Robbins
Abstract
It is a deep-seated tradition in science to view uncertainty as a province of probabilitytheory. The generalized theory of uncertainty (GTU) which is outlined in this paperbreaks with this tradition and views uncertainty in a much broader perspective.Uncertainty is an attribute of information. A fundamental premise of GTU is thatinformation, whatever its form, may be represented as what is called a generalized con-straint. The concept of a generalized constraint is the centerpiece of GTU. In GTU, aprobabilistic constraint is viewed as a special––albeit important––instance of a general-ized constraint.A generalized constraint is a constraint of the form
isr
R
, where
is the con-strained variable,
R
is a constraining relation, generally non-bivalent, and
r
is an index-ing variable which identifies the modality of the constraint, that is, its semantics. Theprincipal constraints are: possibilistic (
r
= blank); probabilistic (
r
=
p
); veristic (
r
=
v
);usuality (
r
=
u
); random set (
r
=
rs
); fuzzy graph (
r
=
fg 
); bimodal (
r
=
bm
); and group
0020-0255/$ - see front matter
Ó
2005 Elsevier Inc. All rights reserved.doi:10.1016/j.ins.2005.01.017
*
Tel.: +1 510 642 4959; fax: +1 510 642 1712.
E-mail address:
zadeh@eecs.berkeley.eduInformation Sciences 172 (2005) 1–40www.elsevier.com/locate/ins
 
(
r
=
). Generalized constraints may be qualified, combined and propagated. The set of all generalized constraints together with rules governing qualification, combination andpropagation constitutes the generalized constraint language (GCL).The generalized constraint language plays a key role in GTU by serving as a precis-iation language for propositions, commands and questions expressed in a natural lan-guage. Thus, in GTU the meaning of a proposition drawn from a natural language isexpressed as a generalized constraint. Furthermore, a proposition plays the role of acarrier of information. This is the basis for equating information to a generalizedconstraint.In GTU, reasoning under uncertainty is treated as propagation of generalized con-straints, in the sense that rules of deduction are equated to rules which govern propaga-tion of generalized constraints. A concept which plays a key role in deduction is that of aprotoform (abbreviation of prototypical form). Basically, a protoform is an abstractedsummary––a summary which serves to identify the deep semantic structure of the objectto which it applies. A deduction rule has two parts: symbolic––expressed in terms of protoforms––and computational.GTU represents a significant change both in perspective and direction in dealing withuncertainty and information. The concepts and techniques introduced in this paper areillustrated by a number of examples.
Ó
2005 Elsevier Inc. All rights reserved.
1. Introduction
Uncertainty is an attribute of information. The path-breaking work of Shannon has led to a universal acceptance of the thesis that information is sta-tistical in nature. A logical consequence of this thesis is that uncertainty, what-ever its form, should be dealt with through the use of probability theory. Toquote an eminent Bayesian, Professor Dennis Lindley, ‘‘The only satisfactorydescription of uncertainty is probability. By this I mean that every uncertaintystatement must be in the form of a probability; that several uncertainties mustbe combined using the rules of probability; and that the calculus of probabil-ities is adequate to handle all situations involving uncertainty
. . .
probability isthe only sensible description of uncertainty and is adequate for all problemsinvolving uncertainty. All other methods are inadequate
. . .
anything that canbe done with fuzzy logic, belief functions, upper and lower probabilities, orany other alternative to probability can better be done with probability’’ (Lind-ley, 1987).The generalized theory of uncertainty (GTU) is a challenge to the thesis andits logical consequence. Basically, GTU puts aside the thesis and its logical con-sequence, and adopts a much more general conceptual structure in which sta-tistical information is just one––albeit an important one––of many forms of information. More specifically, the principal premise of GTU is that, funda-mentally, information is a generalized constraint on the values which a variable
2
L.A. Zadeh / Information Sciences 172 (2005) 1–40
 
is allowed to take. The centerpiece of GTU is the concept of a generalized con-straint––a concept drawn from fuzzy logic, as will be described in greater detailin the sequel. The distinguishing feature of fuzzy logic is that in fuzzy logiceverything is––or is allowed to be––a matter of degree. The principal toolswhich GTU draws from fuzzy logic include precisiated natural language(PNL) and protoform theory (PFT)[56].In GTU, uncertainty is linked to information through the concept of gran-ular structure––a concept which plays a key role in human interaction with thereal world[17,43,52].Informally, a granule of a variable
is a clump of values of 
which aredrawn together by indistinguishability, equivalence, similarity, proximity orfunctionality. For example, an interval is a granule. So is a fuzzy interval.And so is a bell-shaped probability distribution.Granulation is pervasive in human cognition. For example, the granules of age are fuzzy sets labeled young, middle-aged and old,Fig. 1. The granules of height may be very short, short, medium, tall, and very tall. And the granulesof truth may be not true, quite true, not very true, very true, etc. The concept of granularity underlies the concept of a linguistic variable––a concept which wasintroduced in my 1973 paper ‘‘Outline of A New Approach to the Analysis of Complex Systems and Decision Processes’’[41,42]. The concept of a linguisticvariable plays a pivotal role in almost all applications of fuzzy logic[12,15,18,29,31,38].There are four basic rationales which underlie granulation of attributes andthe concomitant use of linguistic variables. First, the bounded ability of sen-sory organs, and ultimately the brain, to resolve detail and store information.For example, looking at Monika, I see that she is young but cannot pinpointher age as a single number. Second, when numerical information may not beavailable. For example, I may not know exactly how many Spanish restaurantsthere are in San Francisco, but my perception may be ‘‘not many.’’ Third,when an attribute is not quantifiable. For example, we describe degrees of 
Age
0121 130
Yearsµ 01Middle-agedYoungOldµ QualizationGranulation (fuzzy quantization)
Fig. 1. Granulation and quantization of age.
L.A. Zadeh / Information Sciences 172 (2005) 1–40
3

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->