Echo State Network with Dung Beetle Optimization Algorithm and Its Application

The echo state network (ESN) has a special hidden layer called the reservoir. Compared with the traditional recurrent neural network (RNN), the ESN has a higher training efficiency, solving the problems of vanishing or exploding gradients. However, there are many parameters in ESN, which can affect the prediction effect. The Dung Beetle Optimization (DBO) algorithm shows great advantages in optimization problems and has a good ability to jump out of the local optimum. In this paper, the DBO is used to optimize the parameters of ESN. Optimizing parameters include reservoir neurons, spectral radius, input scale, and sparsity. The DBO-ESN is applied to the classification of breast cancer, and compares with the ESN based on particle swarm optimization (PSO-ESN), the ESN based on whale optimization algorithm (WOA-ESN), and ESN. The results show that the DBO-ESN has the best classification accuracy.