Successive approximation training algorithm for Feedforward neural networks

摘要

A novel algorithm based on successive approximation training for feedforward neural networks is presented in this paper. The convergence of the algorithm is analysed theoreti- cally and the training error is estimated. Theoretical analysis shows that the novel training algorithm is able to overcome the stalemate problem in the later training stage of the tra- ditional algorithms. Numerical experiments show that the proposed algorithm increases the rate of convergence and improves the generalization performance by avoiding local minima.