Hybrid evolutionary approach for designing neural networks for classification
Zheng‐Hua Tan · Electronics Letters · 2004
An approach for the automatic design of artificial neural networks is presented where a hybrid evolutionary algorithm (HEA) is applied to the structural and parametric learning of networks. The HEA combines genetic algorithms and evolutionary programming on the basis of a real-valued multi-matrix representation. Experimental results show that the proposed approach has a good generalisation and a low computational cost.