Cost-Effective Active Learning for Deep Image Classification

被引:447
作者
Wang, Keze [1 ,2 ]
Zhang, Dongyu [1 ,2 ]
Li, Ya [3 ]
Zhang, Ruimao [1 ,2 ]
Lin, Liang [1 ,2 ]
机构
[1] Sun Yat Sen Univ, Sch Data & Comp Sci, Guangzhou 510006, Guangdong, Peoples R China
[2] Natl Univ Def Technol, Collaborat Innovat Ctr High Performance Comp, Changsha 410073, Hunan, Peoples R China
[3] Guangzhou Univ, Guangzhou 510182, Guangdong, Peoples R China
基金
中国国家自然科学基金;
关键词
Active learning (AL); deep neural nets; image classification; incremental learning;
D O I
10.1109/TCSVT.2016.2589879
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Recent successes in learning-based image classification, however, heavily rely on the large number of annotated training samples, which may require considerable human effort. In this paper, we propose a novel active learning (AL) framework, which is capable of building a competitive classifier with optimal feature representation via a limited amount of labeled training instances in an incremental learning manner. Our approach advances the existing AL methods in two aspects. First, we incorporate deep convolutional neural networks into AL. Through the properly designed framework, the feature representation and the classifier can be simultaneously updated with progressively annotated informative samples. Second, we present a cost-effective sample selection strategy to improve the classification performance with less manual annotations. Unlike traditional methods focusing on only the uncertain samples of low prediction confidence, we especially discover the large amount of high-confidence samples from the unlabeled set for feature learning. Specifically, these high-confidence samples are automatically selected and iteratively assigned pseudolabels. We thus call our framework cost-effective AL (CEAL) standing for the two advantages. Extensive experiments demonstrate that the proposed CEAL framework can achieve promising results on two challenging image classification data sets, i.e., face recognition on the cross-age celebrity face recognition data set database and object categorization on Caltech-256.
引用
收藏
页码:2591 / 2600
页数:10
相关论文
共 42 条
[31]  
Nair V., 2010, PROC INT C MACH LEAR, P807, DOI DOI 10.5555/3104322.3104425
[32]   ImageNet Large Scale Visual Recognition Challenge [J].
Russakovsky, Olga ;
Deng, Jia ;
Su, Hao ;
Krause, Jonathan ;
Satheesh, Sanjeev ;
Ma, Sean ;
Huang, Zhiheng ;
Karpathy, Andrej ;
Khosla, Aditya ;
Bernstein, Michael ;
Berg, Alexander C. ;
Fei-Fei, Li .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 2015, 115 (03) :211-252
[33]  
Scheffer T., 2001, Advances in Intelligent Data Analysis. 4th International Conference, IDA 2001. Proceedings (Lecture Notes in Computer Science Vol.2189), P309
[34]  
Settles B., 2009, TECH REP
[35]   A MATHEMATICAL THEORY OF COMMUNICATION [J].
SHANNON, CE .
BELL SYSTEM TECHNICAL JOURNAL, 1948, 27 (04) :623-656
[36]  
Simonyan K., 2014, 14091556 ARXIV, DOI DOI 10.1016/J.INFSOF.2008.09.005
[37]   Support vector machine active learning with applications to text classification [J].
Tong, S ;
Koller, D .
JOURNAL OF MACHINE LEARNING RESEARCH, 2002, 2 (01) :45-66
[38]   Active Learning Methods for Remote Sensing Image Classification (vol 47, pg 2202, 2009) [J].
Tuia, Devis ;
Ratle, Frederic ;
Pacifici, Fabio ;
Kanevski, Mikhail F. ;
Emery, William J. .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2010, 48 (06) :2767-2767
[39]  
Vijayanarasimhan S, 2011, PROC CVPR IEEE, P1449, DOI 10.1109/CVPR.2011.5995430
[40]   Supervised Descent Method and its Applications to Face Alignment [J].
Xiong, Xuehan ;
De la Torre, Fernando .
2013 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2013, :532-539