Combining Classification and Temporal Learning

Matthew Winston Mitchell, Monash University

This paper introduces TRACA (Temporal Reinforcement-learning and Classification Architecture). TRACA is a connectionist system designed to learn incrementally in environments with irrelevant attributes and hidden-state. A unique method of representing NOT and XOR is used to reduce the complexity of the internal structures developed. TRACA’s network representation is created dynamically during learning avoiding requirements to predetermine network size and topology.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.