点击次数：

论文类型：期刊论文

发表时间：2014-03-27

发表刊物：International Workshop of Extreme Learning Machines (ELM)

收录刊物：SCIE、EI、CPCI-S、Scopus

卷号：128

页面范围：113-118

ISSN号：0925-2312

关键字：DPFNN; ELM; L-1/2 regularizer

摘要：A learning scheme based on Extreme Learning Machine (ELM) and Lip regularization is proposed for a double parallel feedforward neural network. ELM has been widely used as a fast learning method for feedforward networks with a single hidden layer. A key problem for ELM is the choice of the (minimum) number of the hidden nodes. To resolve this problem, we propose to combine the L-1/2 regularization method, that becomes popular in recent years in informatics, with ELM. It is shown in our experiments that the involvement of the L-1/2 regularizer in DPFNN with ELM results in less hidden nodes but equally good performance. (C) 2013 Elsevier B.V. All rights reserved.