ABSTRACT
The Single hidden Layer Feed-forward Neural network (SLFN) has been widely used in many fields for its good performance. The ELM is a new SLFN algorithm (Rongfu Mao, 2009, Guangbin Huang, 2006, Huang, 2005). By using the ELM algorithm, the connection weights between the input layers and the hidden layers, and threshold in the hidden layer, are generated randomly. During the network training, one only needs to set the number of the hidden layer for a unique optimal solution. Compared with the traditional training methods, the ELM method has advantages of learning speed and good generalisation performance. The typical SLFN structure is given in Figure 1, where n denotes the input variables corresponding to n neurons, and m denotes the output variables corresponding to m neurons.
