An Enhanced Fuzzy Min-Max Neural Network Based on Pruning Algorithm for Pattern Classification

Abstract: Various neural networks has been studied based on Fuzzy Min-Max (FMM) for classification of patterns. An Enhanced Fuzzy Min-Max (EFMM) neural network is recent one. The contribution of EFMM is ability to overcome a number of limitations of the original FMM network and increases the performance. The key contributions are three heuristic rules to enhance the learning algorithm of FMM. First, a new rule of hyperbox expansion is introduced to eliminate the overlapping problem which occur while process of hyperbox expansion. Second, other possible overlapping cases are discovered by using newly introduced hyperbox overlap rule. Third, a new hyperbox contraction rule to resolve possible overlapping cases is provided. Due to these rules, EFMM is more complex than FMM. Hence a concept of pruning is introduced in this paper. Confidence factor is used for pruning the hyperboxes and confidence factor is baes of frequency of use and its accuracy of recognition. Pruning of hyperboxes is done to minimize the hyperboxes and improve the network complexity of EFMM, but the recognition rate decreases slightly.

Keywords: Artificial Neural Network, Fuzzy Min-Max Neural Network, Enhanced Fuzzy Min-Max Neural Network, Pruning