Fuzzy Learning Design for Neural Networks

We design a new fuzzy learning mechanism which adaptively teaches the network to acquire a desired pattern as one of stable equilibrium point. The learning mechanism bases on a fuzzy system which combines heuristic learning rules into the design of adaptive learning law, where the heuristic rules are in the form of fuzzy IF-THEN rules. By using Lyapunov synthesis approach, we attain adaptive learning laws adjusting free parameters during learning process. Finally, computer simulation is performed to show the usefulness and effectiveness of our learning mechanism.