[AISWorld] Webinar -- The Possibility Calculus -- July 14, 2016
Sidney Thomas
sthomas at fuzzastat.com
Mon Jun 20 10:23:21 EDT 2016
Dear Colleague:
This is to invite your participation at an introductory seminar/tutorial
on the possibility calculus.
The seminar is to be conducted _Monday July 14, 2016 at 1030h - 1200h
EST_ via webinar. You may participate from the convenience of your home
or office via internet web browser.
There is a charge of _US$199_. To register, reply to this email giving
minimally your name and email address.
This will be the first of a series of these seminar/tutorials to be
offered twice a week, Mondays and Wednesdays, until October 3, 2016,
with a break taken the first week (Labor Day week) of September. Links
are given below for the course syllabus among others.
The first seminar/tutorial will give an overview of the entire course,
and in particular, it will seek to motivate and elucidate the central
conceptual breakthrough that is at the bottom of the possibility calculus.
Why should you be interested in this course of seminars?
* it addresses and solves the problem at the foundations of
statistical inference, namely the rules that ought to apply for
drawing inference from the (absolute) likelihood function. In place
of the maximization rule of Fisher, and the integration rule (in
effect) of the Bayesians, it is shown how the correct rule of
disjunction should always have been a rule of product-sum, and
deriving therefrom the notion of the product-sum integral. This also
means that likelihood should always have been defined in an
absolute, not relative sense. Admittedly, this is difficult without
the idea of data being fuzzy in general;
* it proposes a new criterion for regression analysis in line with the
foregoing. Maximizing the probability of the data translates into
maximizing a product-sum integral. In some circumstances this yields
answers intriguingly different from the MLE and Bayesian criteria; and
* it proposes a reformulated fuzzy logic (FL) in which the AND and OR
connectives are term-functional rather than truth-functional. As a
result, it is possible to uphold the Aristotelian rules of form
where these are applicable, without sacrificing the fuzziness, and
without sacrificing the central triumph of the truth-functional FL
in its application to control problems and approximate reasoning. It
becomes possible also to apply the reformulated FL to the problem of
statistical inference without misgiving. For the problem of
statistical inference is a problem in FL to the extent it is the
problem of characterizing what the data -- fuzzily-- say about the
uncertain parameter constants.
A free pdf copy of the monograph, /The Possibility Calculus: The Unified
Theory of Fuzzy Logic, The Possibility Calculus, and Statistical
Inference /(249 pp.) will be included with your attendance. The
monograph advances 72 new theorems and lemmas. It is a serious
contribution to the field.
I hope you will attend.
Kind regards
Sidney
--
Sidney Thomas (Ph.D., 1980, University of Toronto, Operations Research)
Author: The Possibility Calculus (2016), Fuzziness and Probability (1995)
_Links:_
Seminar info: http://fuzzastat.com/seminar.html
Course syllabus: http://fuzzastat.com/PossSyllabus.pdf
Book excerpts: http://fuzzastat.com/excerpts.html
More information about the AISWorld
mailing list