Reconfigurable neural nets by energy convergence learning principle based on extended McCulloch-Pitts neurons and synapses

An energy landscape approach to designing neural nets is simple and powerful. The nature of competitive and cooperative learning is similar to that studied by S. Grossberg et al. (1976) and the D. Rumelhart PDP school, but differs slightly in the principles and neuronic models used. This model of hairy neurons emphasizes an active growth role played by peripheral neurofilaments in neural net computing which cannot be solely attributed to the neuronic core matter because of a neurochemical independence. Although protein acting forces guide neurite growth and synapse formation, neuronic firing rates are responsible for synaptic efficacies.<>