Sparse Bayesian approach to fast learning network for multiclassification

Abstract

This paper proposes a novel artificial neural network called sparse-Bayesian–based fast learning network(SBFLN). In SBFLN, sparse Bayesian regression is used to train the fast learning network (FLN), which is an improvedextreme learning machine (ELM). The training process of SBFLN is to randomly generate the input weights and thehidden layer biases, and then find the probability distribution of other weights by the sparse Bayesian approach. SBFLNcalculates the predicted output through Bayes estimator, so it can provide a natural marginal possibility for classificationproblems and can solve the overfitting problem caused by the least-squares estimation in FLN. In addition, the sparseBayesian approach can automatically trim most redundant neurons in hidden layer, which makes the network morecompact and accurate. To verify the effectiveness of the improvements in this paper, the results of SBFLN are evaluatedin 15 benchmark classification problems. The experimental results show that SBFLN is not sensitive to the number ofneurons in the hidden layer, and the performance of SBFLN is competitive or superior to some other state-of-the-artalgorithms.