Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!news.moneng.mei.com!uwm.edu!msuinfo!harbinger.cc.monash.edu.au!bruce.cs.monash.edu.au!lloyd
From: lloyd@cs.monash.edu.au (Lloyd Allison)
Subject: Re: Induction of finite state machines
Message-ID: <lloyd.784852442@bruce.cs.monash.edu.au>
Sender: news@bruce.cs.monash.edu.au (USENET News System)
Organization: Computer Science, Monash University, Australia
References: <39ui1i$t6@pollux.usc.edu>
Date: Mon, 14 Nov 1994 22:34:02 GMT
Lines: 78

skowalsk@pollux.usc.edu (Stephen Kowalski) writes:

>I am conducting a survey on the induction of finite state machines and
>would appreciate any references.  

>Thanks,
>Steve Kowalski
>University of Southern California


W&B worked out the information-theory of the multi-state distribution in 1968,
which gives you the edge transitions *if you know the FSM architecture*
or if you are comparing some given FSM architectures.

%A D. M. Boulton
%A C. S. Wallace
%T The information content of a multistate distribution
%J J. Theor. Biol.
%V 23
%P 269-278
%D 1969
%X see also Wallace & Boulton COMPJ (Computer Journal) 1968


We have used this to compare a few different architectures to
model DNA evolution, eg:

%A L. Allison
%A C. S. Wallace
%A C. N. Yee
%T Finite-state models in the alignment of macro-molecules.
%J J. Mol. Evol.
%V 35
%N 1
%P 77-89
%M JUL
%D 1992


FSM are similar to hidden Markov models, much used in speech recognition,
compression, and protein comparison, eg.:

%A David Haussler
%A Anders Krogh
%A Saira Mian
%A Kimmen Sjolander
%T Protein modeling using hidden Markov models: analysis of globins.
%R UCSC-CRL-92-23
%I University of California at Santa Cruz,
%M JUN
%D 1992


The difficult bit with FSMs is inferring the architecture (unless
you have just a few likely candidates).  Fogel has tried it for IPD:

%A D. B. Fogel
%T Evolving behaviours in the iterated prisoner's dilemma.
%J Evolutionary Computation.
%V 1
%N 1
%P 77-97
%D 1993


More information and refs are available via the www:
<A HREF="http://www.cs.monash.edu.au/~lloyd/tildeMML">MML, Inductive Inf'</A>
<A HREF="http://www.cs.monash.edu.au/~lloyd/tildeStrings">Molecular Biology</A>
<A HREF="http://www.cs.monash.edu.au/~lloyd/tildeBIB">Bibliography</A>


Hope you will post a summary.

Lloyd ALLISON
Central Inductive Agency,
Dept. of Computer Science, Monash University, Clayton, Victoria 3168, AUSTRALIA
tel: 61 3 905 5205       fax: 61 3 905 5146       email: lloyd@cs.monash.edu.au

