LGIST: Learning Generalized Image Schemas for Transfer
Abstract
This research was funded under a Thrust D award to develop algorithms for gist memory. Gists are abstractions that preserve the important causal events in activities or episodes, eliding details. The point of this research was to develop a new kind of representation, image schemas, and associated learning methods. The learning agent was called Jean. Jean is a model of early cognitive development based loosely on Jean Piaget's theory of sensori-motor and pre-operational thought. Like an infant, Jean repeatedly executes schemas, gradually transferring them to new situations and extending them as necessary to accommodate new experiences. We model this process of accommodation with the Experimental State Splitting (ESS) algorithm. ESS learns elementary action schemas, which comprise controllers and maps of the expected dynamics of executing controllers in different conditions. ESS also learns compositions of action schemas called gists. We present tests of the ESS algorithm in three transfer learning experiments, in which Jean transfers learned gists to new situations in a real time strategy military simulator.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2008
- Accession Number
- ADA491488
Entities
People
- Carole Beal
- Paul Cohen
Organizations
- University of Southern California