New Approach for Interpretability of Neuro-Fuzzy Systems with Parametrized Triangular Norms

In this paper we proposed a new approach for interpretability of the neuro-fuzzy systems. It is based on appropriate use of parametric triangular norms with weights of arguments, which shape depends on values of their parameters and weights. The use of those norms as aggregation and inference operators increases precision of fuzzy system. Due to that, the rule base can be simpler and easier to interpretation. However, interpretation of parametric triangular norms is not that obvious as interpretation of nonparametric triangular norms such as algebraic or minimal norms. Proposed approach is based on choosing values of parameters from a set of values, where each value have its own interpretation. Additionally, a modified tuning algorithm for selection both the structure and structure parameters of fuzzy system with interpretability criteria under consideration is proposed. Proposed approach were tested on well-known nonlinear simulation problems.