Input selection and rule generation in adaptive neuro-fuzzy inference system for protein structure prediction

Abstract

This work presents a method for input selection based on fuzzy clustering and for rule generation based on genetic algorithm (GA) in an adaptive neuro-fuzzy inference system (ANFIS) which is used for modeling protein secondary structure prediction. A two-phase process is employed in the model. In the first phase, the selection of number and position of the fuzzy sets of initial input variables can be determined by employing a fuzzy clustering algorithm; and in the second phase the more precise structural identification and optimal parameters of the rule-base of the ANFIS are achieved by an iterative GA updating algorithm. An experiment on three-state secondary structure prediction of protein is reported briefly and the performance of the proposed method is evaluated. The results indicate an improvement in design cycle and convergence to the optimal rule-base within a relatively short period of time, at the cost of little decrease in accuracy. © Springer-Verlag 2003.