A SELF-ADJUSTING THE NUMBER OF HIDDEN LAYER NEURON ALGORITHM BASED ON BP NETWORK

摘要

The number of hidden-layer neurons in the BP network has drawn tremendous attention from both academia and industry. This paper proposes a self-adaptive algorithm for the AB network model with a series of new mechanisms. Our simulation experiments demonstrate that this algorithm improve performance significantly, in terms of hidden-layer neutrons, as well as the rate of convergence compared with the classical BP algorithm.