Path: utzoo!attcan!uunet!lll-winken!lll-tis!helios.ee.lbl.gov!pasteur!agate!bionet!apple!bloom-beacon!tut.cis.ohio-state.edu!husc6!linus!mbunix!bwk
From: bwk@mitre-bedford.ARPA (Barry W. Kort)
Newsgroups: comp.ai
Subject: Re: Valiant's Learning Model
Summary: Some classical learning models.
Keywords: Computational Learning Theory
Message-ID: <41644@linus.UUCP>
Date: 8 Nov 88 13:44:38 GMT
References: <6083@techunix.BITNET>
Sender: news@linus.UUCP
Reply-To: bwk@mbunix (Kort)
Organization: The MITRE Corporation, Bedford, Mass.
Lines: 13

In article <6083@techunix.BITNET> dario@techunix.BITNET (Dario Ringach) writes:

 > Has anyone attempted to approach learning as a discrete time Markov
 > process on the hypothesis space H?  For instance at any time k let
 > h1=h(k) be the current hypothesis obviously there is defined for any
 > h2 in H a transition probability P(h(h+1)=h2|h(k)=h1) that depends
 > on the probability distribution Px and the learning algorithm A.

Look into Bayesian inference, Kalman filtering, and Kailath's
Innovations Process.  In each of these approaches, a current
best guess is updated as new information comes in.  I believe
Widrow's adaptive networks also exhibit such behavior.

--Barry Kort