文献类型：期刊文献

英文题名：Extreme learning machines for regression based on V-matrix method

作者：Yang, Zhiyong[1,2];Zhang, Taohong[1,2];Lu, Jingcheng[1];Su, Yuan[3];Zhang, Dezheng[1,2];Duan, Yaowu[4]

第一作者：Yang, Zhiyong

通讯作者：Zhang, DZ[1];Zhang, DZ[2]

机构：[1]Univ Sci & Technol Beijing, Sch Comp & Commun Engn, Dept Comp, Beijing 100083, Peoples R China;[2]Beijing Key Lab Knowledge Engn Mat Sci, Beijing 100083, Peoples R China;[3]Univ Maryland, Joint Ctr Quantum Informat & Comp Sci, College Pk, MD 20742 USA;[4]Beijing Union Univ, Dept Basic Course, Biochem Engn Coll, Beijing 100023, Peoples R China

第一机构：Univ Sci & Technol Beijing, Sch Comp & Commun Engn, Dept Comp, Beijing 100083, Peoples R China

通讯机构：[1]corresponding author), Univ Sci & Technol Beijing, Sch Comp & Commun Engn, Dept Comp, Beijing 100083, Peoples R China;[2]corresponding author), Beijing Key Lab Knowledge Engn Mat Sci, Beijing 100083, Peoples R China.

年份：2017

卷号：11

期号：5

起止页码：453-465

外文期刊名：COGNITIVE NEURODYNAMICS

收录：;Scopus(收录号：2-s2.0-85020647319);WOS:【SCI-EXPANDED(收录号:WOS:000410473600005)】；

基金：This paper is supported by the Scientific Research Foundation for the Returned Overseas Chinese Scholars, National Key Technology RD Program in 12th Five-year Plan of China (No. 2013BAI13B06) and National Key R&D Plan, Cloud Computing and Big Data Special Plan, 2017.

语种：英文

外文关键词：Extreme learning machine; V matrix; Regression

摘要：This paper studies the joint effect of V-matrix, a recently proposed framework for statistical inferences, and extreme learning machine (ELM) on regression problems. First of all, a novel algorithm is proposed to efficiently evaluate the V-matrix. Secondly, a novel weighted ELM algorithm called V-ELM is proposed based on the explicit kernel mapping of ELM and the V-matrix method. Though V-matrix method could capture the geometrical structure of training data, it tends to assign a higher weight to instance with smaller input value. In order to avoid this bias, a novel method called VI-ELM is proposed by minimizing both the regression error and the V-matrix weighted error simultaneously. Finally, experiment results on 12 real world benchmark datasets show the effectiveness of our proposed methods.