Gradient Projected Neuro Fuzzy Decision Tree

Fuzzy Decision Tree induction is a powerful methodology to extract human interpretable classification rules. The induction of fuzzy decision tree is done using popular Fuzzy ID3 algorithm [1, 2]. The other parameters which influence the performance of fuzzy decision tree are cut standard parameter  and leaf selection threshold parameter th  . Generally these parameters are to be selected heuristically or by trial and error approach. To further improve the performance of Fuzzy decision tree the parameters of FDT are optimized after obtaining the fuzzy rules using fuzzy ID3. This parameter optimization process has improved the accuracy of the FDT in Rajen et.al work [3]. Rajen et.al has proposed a methodology to improve the accuracy by applying back propagation algorithm directly on the FDT structure without compromising the interpretability. The existing results show that there is no control on the growth of parameters during optimization. To control the growth of certain parameters during tuning the concept of gradient projection is used over Neuro Fuzzy Decision tree and the model is named as Gradient Projected Neuro Fuzzy Decision Tree (GP-N-FDT).This methodology can be used to solve the minimization problems with simple bounds.