ABSTRACT
The brief history of neural network research presented in this section indicates that, although the initial revolution in neural networks lost its early momentum, the second revolution may well avoid the fate of the first. The subject now has strengths that were absent from its earliest version: these are discussed, and especially the fact that the biological origin of the subject is now giving it greater stability. The new avenues opened up by biologically motivated research and by studies in other areas such as statistical mechanics, statistics, functional analysis and machine learning are described, and future directions discussed. The strengths and weaknesses of the subject are compared with those of alternative and competing approaches to information processing.
