A T-model neural network with learning ability

The authors introduce a novel neural network called the T-model with learning capability. They utilized the least mean square algorithm to train the T-model and exploit the learning capability of the introduced network. The architecture and the algorithm of the T-model network are described. Some typical examples to illustrate the effectiveness of the algorithm are included and compared with the Hopfield model and a multilayer model. The simulations and experiments showed that the algorithm is quite effective in training the T-model network for many problems.<<ETX>>