|
|
【答案】应助回帖
★ ★ ★ ★ ★
感谢参与，应助指数 +1
baroshi: 金币+5, ★★★★★最佳答案 2014-09-21 19:06:41
sunshan4379: LS-EPI+1, 感谢应助！ 2014-09-21 19:36:07
A compressed sensing approach for efficient ensemble learning
作者:Li, L (Li, Lin)[ 1 ] ; Stolkin, R (Stolkin, Rustam)[ 2 ] ; Jiao, LC (Jiao, Licheng)[ 1 ] ; Liu, F (Liu, Fang)[ 1 ] ; Wang, S (Wang, Shuang)[ 1 ]
PATTERN RECOGNITION
卷: 47
期: 10
页: 3451-3465
DOI: 10.1016/j.patcog.2014.04.015
出版年: OCT 2014
查看期刊信息
摘要
This paper presents a method for improved ensemble learning, by treating the optimization of an ensemble of classifiers as a compressed sensing problem. Ensemble learning methods improve the performance of a learned predictor by integrating a weighted combination of multiple predictive models. Ideally, the number of models needed in the ensemble should be minimized, while optimizing the weights associated with each included model. We solve this problem by treating it as an example of the compressed sensing problem, in which a sparse solution must be reconstructed from an under-determined linear system. Compressed sensing techniques are then employed to find an ensemble which is both small and effective. An additional contribution of this paper, is to present a new performance evaluation method (a new pairwise diversity measurement) called the roulette-wheel kappa-error. This method takes into account the different weightings of the classifiers, and also reduces the total number of pairs of classifiers needed in the kappa-error diagram, by selecting pairs through a roulette-wheel selection method according to the weightings of the classifiers. This approach can greatly improve the clarity and informativeness of the kappa-error diagram, especially when the number of classifiers in the ensemble is large. We use 25 different public data sets to evaluate and compare the performance of compressed sensing ensembles using four different sparse reconstruction algorithms, combined with two different classifier learning algorithms and two different training data manipulation techniques. We also give the comparison experiments of our method against another five state-of-the-art pruning methods. These experiments show that our method produces comparable or better accuracy, while being significantly faster than the compared methods. (C) 2014 Elsevier Ltd. All rights reserved.
关键词
作者关键词:Ensemble learning; Classification; Classifier ensemble; Sparse reconstruction; Compressed sensing; Roulette-wheel selection; Kappa-error
KeyWords Plus:RANDOM SUBSPACE METHOD; CLASSIFIER ENSEMBLES; NEURAL-NETWORKS; ALGORITHMS; REGRESSION; DIVERSITY
作者信息
通讯作者地址: Li, L (通讯作者)
[显示增强组织信息的名称] Xidian Univ, Int Res Ctr Intelligent Percept & Computat, Minist Educ, Key Lab Intelligent Percept & Image Understanding, Xian 710071, Peoples R China.
地址:
[显示增强组织信息的名称] [ 1 ] Xidian Univ, Int Res Ctr Intelligent Percept & Computat, Minist Educ, Key Lab Intelligent Percept & Image Understanding, Xian 710071, Peoples R China
[显示增强组织信息的名称] [ 2 ] Univ Birmingham, Sch Mech Engn, Birmingham B15 2TT, W Midlands, England
电子邮件地址:xdlinli86@163.com; lchjiao@mail.xidian.edu.cn
基金资助致谢
基金资助机构 授权号
National Natural Science Foundation of China
61371201
61001202
61272279
61273317
National Top Youth Talents Support Program of China
查看基金资助信息
出版商
ELSEVIER SCI LTD, THE BOULEVARD, LANGFORD LANE, KIDLINGTON, OXFORD OX5 1GB, OXON, ENGLAND
类别 / 分类
研究方向:Computer Science; Engineering
Web of Science 类别:Computer Science, Artificial Intelligence; Engineering, Electrical & Electronic
文献信息
文献类型:Article
语种:English
入藏号: WOS:000338392400020
ISSN: 0031-3203
电子 ISSN: 1873-5142
期刊信息
目录： Current Contents Connect®
Impact Factor (影响因子): Journal Citation Reports®
其他信息
IDS 号: AK4KA
Web of Science 核心合集中的 "引用的参考文献": 47
Web of Science 核心合集中的 "被引频次": 0 |
|
回复此楼