Development of Discrete-Cockroach Algorithm (DCA) for Feature Selection Optimization

Abstract

One of the recently proposed algorithms in the field of bio-inspired algorithm is the Hungry Roach Infestation Optimization (HRIO) algorithm. Haven has developed optimization algorithms HRIO that is inspired by recent discoveries in the social behavior of cockroaches. Result showed that HRIO was effective at finding the global optima of a suite of test functions. However, there is no researcher who has observed HRIO for solving discrete feature selection problems. Therefore, we try to develop a discrete-cockroach algorithm (DCA) as the modification of HRIO for optimizing discrete feature selection problem. We test the algorithm to solve feature selection problem in machine vision for predicting water stress in plant using single and multiobjectives optimization. Two objective functions are minimizing prediction error and minimizing the number of feature-subset. The results showed DCA has better performance compared to the existed bio-inspired optimization algorithms such as genetic algorithms (GA) and discreteparticle swarm optimization (discrete-PSO). The performance showed significant difference between DCA and other methods.