Competitive networks, including self-organizing maps

Abstract

This chapter examines the basic operation of competitive neural networks, including self-organizing maps. It explains that competitive neural networks learn to categorize input pattern vectors and each category of inputs activates a different output neuron. This categorization is of great potential importance in perceptual systems because each category formed reflects a set or cluster of active inputs. The chapter provides a diagram showing the basic architecture of a competitive network, which is a one-layer network with a set of inputs that make modifiable excitatory synapses with the output neurons.