OPTIMIZING NEURAL NETWORK CLASSIFIER FOR DIABETES DATA USING METAHEURISTIC ALGORITHMS

Abstract

Abstract – This paper reports the performance comparison among several metaheuristics algorithms on the neural network training. In this research we use five metaheuristic algorithms which implements for diabetes data, there are Particle Swarm Optimizer (PSO), Multi-Verse Optimizer (MVO), Grey Wolf Optimizer (GWO), Bat Algorithm (BAT), and Cuckoo Search (CS). The Cuckoo Search (CS) algorithm is a recently developed meta-heuristic optimization algorithm which is suitable for solving optimization problems. The main problem to be solved is to find the most effective meta-heuristic optimization algorithm. The search was done by comparing the results of PSO (Particle Swarm Optimizer) algorithm test with the test with MVO (Multi-Verse Optimizer), GWO (Grey Wolf Optimizer), BAT (Bat Algorithm), and CS (Cuckoo Search). Then look for the most effective algorithm. The best metaheuristic algorithm that we had in this research is MVO, with best case accuracy result 78% and lowest standard deviation is 0.00675 and the worst is BAT algorithm with best case accuracy 77% and standard deviation 0.14571.