Non-uniform self-selective coder for fuzzy rules and its application

摘要

We reconsider the application scope of fuzzy rules and find the following. (1) Using spline, we can easily obtain more accurate results than those obtain by the generalized dynamic fuzzy neural network. (2) If the model is nonlinear with a disturbance term, we obtain that the checking error is very large even though the training error is small. If the model is chaotic with a disturbance term, we obtain that both the training error and checking error are very large. (3) Using a sequential algorithm as in the generalized dynamic fuzzy neural network, we would always be trapped at the local minima rather than the global minimum. Therefore we use the non-uniform self-selective coder instead and show how it works by an empirical example.