Evolving neural network with extreme learning for system modeling

This p aper introduces an evolving feedforward single hidden layer neural network with extreme learning. The evolving neural network simultaneously adapts its structure and updates its weights using recursive algorithms. Neurons in the hidden layer are added whenever necessary by the implicit nature of the input data. The number of neurons in the hidden layer is found using a recursive granulation algorithm based on the concept of cloud. A cloud is a collection of points whose density implicitly defines a cluster. An extreme learning-based algorithm is used to compute hidden and output layers weights of the neural network. Computational results show that the evolving neural network modeling approach is competitive when compared with alternative evolving modeling approaches.