A fast algorithm for training a class of fuzzy neural networks
Dongmei Li, Liu Junqiang, Hu Hengzhang · 2002
A fast algorithm for training a class of fuzzy neural networks (FNN) is studied. The proposed algorithm is called least square-simplex (LS-simplex). The algorithm obtains the performance of global convergence and avoids the inherent local convergence when adopting a grads algorithm to train the FNN, also it accelerates the FNN's training and can be used online which is impossible when using a genetic algorithm (GA). Compared with the grads algorithm and GA, the LS-simplex owns more accurate precision and faster convergent speed, and the FNN obtained has excellent generalization performance.