Extreme Learning Machine
Author: Kailugaji - blog Park http://www.cnblogs.com/kailugaji/
2004 Nanyang Technological University Huangguang Bin ELM proposed algorithm. Extreme Learning Machine (ELM Extreme Learning Machine) is a fast before a single hidden layer feedforward neural network (SLFN) training algorithm.
Characteristics of the algorithm is the process of determining the network parameters, the hidden layer parameters (a, b) randomly selected, no adjustment in the training process, only need to set the number of hidden layer neurons, it is possible to obtain the single most an optimum solution; while the right outer (i.e., the output weights) network least squares solution loss function obtained by minimizing the squares (eventually normalized to solving of Moore-Penrose generalized inverse matrix of a problem) such network parameter determination process in step without any iteration, thereby greatly reducing the time adjusting network parameters. Compared with traditional training methods, this method has fast learning, good generalization performance advantages.
reference:
[2] MATLAB program: ELM speed learning machine
[3] debate Yann LeCun: Can anyone explain Extreme Learning Machine (ELM) cattle X Where?