Grammatical Differential Evolution Adaptable Particle Swarm Optimizer for Artificial Neural Network Training

In this paper, Grammatical Differential Evolution (GDE) Adaptable Particle Swarm Optimizer (GDE-APSO) is applied in training of Feed –Forward Neural Network (FNN). The error function of FNN is a highly multimodal function having too many local optima. GDEAPSO algorithm can solve multimodal function efficiently and effectively. The devised method is termed as GDEAPSO- NN and it is used to train FNN for XOR problem. Here, the CLPSO and DE/best/1/bin algorithms are also applied to train FNN for the same problem to make a comparative study. The experimental study shows that GDEAPSO performed better than CLPSO and DE/best/1/bin algorithm in FNN training.