Cost-conscious comparison of supervised learning algorithms over multiple data sets

作     者：Ulas, Aydin Yildiz, Olcay Taner Alpaydin, Ethem 

作者机构：Bogazici Univ Dept Comp Engn TR-34342 Istanbul Turkey Isik Univ Dept Comp Engn TR-34398 Istanbul Turkey 

出 版 物：《PATTERN RECOGNITION》 (图形识别)

年 卷 期：2012年第45卷第4期

页      面：1772-1781页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Turkish Academy of Sciences [EA-TUBA-GEBIP/2001-1-1] Bogazici University [07HA101] Turkish Scientific Technical Research Council (TUBITAK) [EEEAG 107E127, 107E222, 109E186] 

主　　题：Machine learning Statistical tests Classifier comparison Model selection Model complexity 

摘      要：In the literature, there exist statistical tests to compare supervised learning algorithms on multiple data sets in terms of accuracy but they do not always generate an ordering. We propose Multi(2)Test, a generalization of our previous work, for ordering multiple learning algorithms on multiple data sets from best to worst where our goodness measure is composed of a prior cost term additional to generalization error. Our simulations show that Multi2Test generates orderings using pairwise tests on error and different types of cost using time and space complexity of the learning algorithms. (C) 2011 Elsevier Ltd. All rights reserved.