Optimizing a Multi-Layer Perceptron Based on an Improved Gray Wolf Algorithm to Identify Plant Diseases

Metaheuristic optimization algorithms play a crucial role in problems. However, the traditional identification methods have following problems: (1) difficulties nonlinear data processing; (2) high error rates caused by local stagnation; and (3) low classification resulting from premature convergence. This paper proposed variant based on gray wolf algorithm (GWO) with chaotic disturbance, candidate migration, attacking mechanisms, naming it enhanced optimizer (EGWO), to solve problem of convergence stagnation. The performance EGWO was tested IEEE CEC 2014 benchmark functions, results were compared three GWO variants, five popular algorithms, six recent algorithms. In addition, optimized weights biases multi-layer perceptron (MLP) an EGWO-MLP disease model; model verified UCI dataset including Tic-Tac-Toe, Heart, XOR, Balloon datasets. experimental demonstrate that can effectively avoid problems provide quasi-optimal solution for problem.