Monash U SEMINAR: A decision graph explan'n of protein 2'ry structure

[David L Dowe]

Hello.   Below are the details of a SEMINAR to be given at Monash University,
Melbourne, Australia on Wednesday 16th December.  (Sorry if I've got too many
countries or too many newsgroups.  Please pardon.)  The approach to protein
(secondary) structure prediction is information-theoretic, and involves
machine-learning techniques.   Further details follow.  Thank you.

_________________________________________________________________
Date: Tue, 8 Dec 92 16:42:30 +110
From: karen at bruce.cs.monash.EDU.AU (karen patricia fenwick
Subject: Seminar 16th December >Status: R

            DEPARTMENT OF COMPUTER SCIENCE MONASH UNIVERSITY CLAYTON
                                     SEMINAR
Time:   Wednesday, 16th December, 1992  at 2.15 pm
Place:  Room 135, Computer Science Building

Topic:  A Decision Graph Explanation of Protein Secondary Structure 
        Prediction

Speaker:Dr. David L. Dowe 
        Department of Computer Science,
        Monash University, Clayton, 3168, AUSTRALIA.

Abstract:

Oliver and Wallace recently introduced the machine-learning technique
of decision graphs, a generalisation of decision trees.  Here it is
applied to the prediction of protein secondary structure to infer a
theory for this problem.  The resulting decision graph provides both a
prediction method and, perhaps more interestingly, an explanation for
the problem.  Many decision graphs are possible for the problem;
a particular graph is just one theory or hypothesis of secondary
structure formation.  Minimum message length encoding is used to judge
the quality of different theories.
It is a general technique of inductive inference
and is resistant to learning the noise in the training data.

The method was applied to 75 sequences from non-homologous proteins
comprising 13K amino acids.
The predictive accuracy for 3 states (Extended, Helix, Other) is in the
range achieved by current methods.  Many believe this is close to the
limit for methods based on only local information.  However, a
decision graph is an explanation of a phenomenon.  This contrasts with
some other predictive techniques which are more "opaque".  The reason
for a prediction can be of more value than the prediction itself.
    _________________________________________________________________
For further details, please e-mail (in the first instance)
karen at bruce.cs.monash.edu.au (on behalf of the seminar co-ordinator) or
(the speaker) dld at bruce.cs.monash.edu.au .

Thank you.