Cascading fuzzy neural networks

A new fuzzy neural network (FNN) model based on syllogistic fuzzy reasoning is proposed. Unlike most of the proposed FNN models which implement single-stage fuzzy reasoning mechanisms, the new model implements syllogistic fuzzy reasoning which is usually used by human beings. A hybrid learning algorithm is presented to derive an appropriate syllogistic fuzzy rule set and update the corresponding network parameters. It is found that the model is superior to its single-stage counterpart in learning ability and robustness.