Computational Explorations of the Baldwin Eect
Keith L. Downing · 2009
rst constructive proof of its potential existence, and subsequent work in evolutionary computation has shown the practical, computational, advantages of hybrid evolution-learning systems. However, the basic theory, particularly it’s second phase (involving genetic assimilation of acquired characteristics) is dicult to reconcile in systems controlled by neural networks. This paper gives a brief overview of early B.E. research, describes one of our earlier projects involving B.E. and trilaterally adaptive systems, and sketches our current focus on the investigation of B.E. in trilaterally-adaptive neural networks.