An extreme learning machine model based on adaptive multi-fusion chaotic sparrow search algorithm for regression and classification

Extreme learning machine (ELM) is a machine learning algorithm based on single hidden layer feedforward neural network. Compared with traditional neural network algorithms, ELM has the advantages of fast training speed and good generalization ability with guaranteed learning accuracy, which has been widely used in practical problems. However, due to the randomness of hidden layer parameters, the original ELM requires more hidden neurons in applications, making the network more complex. For this reason, we propose an improved extreme learning machine based on the adaptive multi-fusion chaotic sparrow search algorithm (AMFCSSA), called AMFCSSA-ELM. In our work, AMFCSSA is specially designed which combines the techniques of chaotic mapping, adaptive weighting coefficients, variable spiral factors and t -distribution mutation, to optimize the input weights and hidden biases of ELM. Firstly, unimodal and multimodal test functions are used to verify the performance of AMFCSSA, the results show that AMFCSSA can search for the global optimal value better than PSO, SSA, and so on. Secondly, various nonlinear functions and classification datasets are taken to test the capability of AMFCSSA-ELM, the simulation results show that our method has better generalization performance compared with other similar algorithms such as PSO-ELM, SSA-ELM.