Synergy between Extreme Learning Machine and Gaussian artificial immune system to train multilayer perceptrons

Extreme Learning Machine (ELM) is a novel and fast approach to train single-layer feedforward neural networks. With ELM, the input weights and hidden biases are chosen randomly and the output weights are analytically determined using the inverse operation of the hidden layer output matrices. Due to this randomness of the input weights, ELM generally tends to require a higher number of hidden nodes to achieve a good generalization performance. This paper proposes a hybrid approach to train MLPs which combines the advantages of a powerful artificial immune system, called GAIS, with the advantages of ELM, eliminating the individual limitations of both when applied in isolation to the same task. In this proposal, the GAIS algorithm is responsible for finding a proper set of input weights whereas the output weights are determined by the Moore-Penrose generalized inverse. The proposed methodology was evaluated in seven well-known classification problems and its performance compares favorably with that produced by contenders, such as ELM, E-ELM and opt-aiNet.