Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site ucbvax.BERKELEY.EDU
Path: utzoo!decvax!bellcore!ulysses!ucbvax!XX.LCS.MIT.EDU!JHC%OZ.AI.MIT.EDU
From: JHC%OZ.AI.MIT.EDU@XX.LCS.MIT.EDU
Newsgroups: mod.ai
Subject: Seminar - Learning to Construct Abstractions (MIT)
Message-ID: <JHC.12206640012.BABYL@MIT-OZ>
Date: Wed, 14-May-86 11:16:00 EDT
Article-I.D.: MIT-OZ.JHC.12206640012.BABYL
Posted: Wed May 14 11:16:00 1986
Date-Received: Tue, 27-May-86 17:56:00 EDT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 21
Approved: ailist@sri-ai.arpa


                    -- AI Revolving Seminar --

                LEARNING TO CONSTRUCT ABSTRACTIONS

                          Rick Lathrop
                           MIT AI Lab

One useful trait of an intelligent agent is to construct higher-level
abstractions from a mass of detailed low-level information.  This talk
will explore one way an agent might be taught how to construct such
abstractions, and why it might be a useful or interesting for an agent
to do so.  A main motivation is the possibility of the use of these
abstractions to see similarities (between situations) that are
obscured by the mass of irrelevant details at the lower level.
Preliminary examples from the Rieger (causal) mechanism world, VLSI
circuit analysis, and protein structure analysis will be discussed.


Thursday, May 15, 4pm
NE-43, 8th floor playroom