Evolution and Learning in Neural Networks: Dynamic Correlation, Relearning and Thresholding

Carse, Brian; Oreland, Johan

Abstract

This contribution revisits an earlier discovered observation that the average performance of a pop ulation of neural networks that are evolved to solve one task is improved by lifetime learning on a different task. Two extant, and very different, explanations of this phenomenon are examined- dynamic correlation, and relearning. Experimental results are presented which suggest that neither of these hypotheses can fully explain the phenomenon. A new explanation of the effect is proposed and empirically justified. This explanation is based on the fact that in these, and many other relat ed studies, real-valued neural network outputs are thresholded to provide discrete actions. The effect of such thresholding produces a particular type of fitness landscape in which lifetime learn ing can reduce the deleterious effects of mutation, and therefore increase mean population fitness. © 2000, Sage Publications. All rights reserved.

• Journal Article Type: Article
• Publication Date: Jan 1, 2000
• Journal: Adaptive Behavior
• Print ISSN: 1059-7123
• Electronic ISSN: 1741-2633
• Publisher: SAGE Publications
• Peer Reviewed: Not Peer Reviewed
• Volume: 8
• Issue: 4
• Pages: 297-311
• DOI: https://doi.org/10.1177/105971230000800305
• Keywords: evolution, learning, neural networks, dynamic correlation, relearning, thresholding
• Public URL: https://uwe-repository.worktribe.com/output/1093167
• Publisher URL: http://dx.doi.org/10.1177/105971230000800305
• Additional Information: Additional Information : An earlier version of this paper was originally presented at the Genetic and Evolutionary computational conference (GECCO) in July, 2000. The work explores the interactions between life-long learning and artificial evolution. The work is significant since it critically examines two existing theories of learning/evolution interaction and proposes a new mechanism by which this can occur; namely that lifetime learning can increase an individual's resilience to deleterious mutations during reproduction.