Incremental learning of concept drift in Multiple Instance Learning for industrial visual inspection

被引:13
作者
Mera, Carlos [1 ]
Orozco-Alzate, Mauricio [2 ]
Branch, John [3 ]
机构
[1] Inst Tecnol Metropolitano Medellin, Dept Sistemas Informac, Calle 54A 30-01, Medellin 050013, Colombia
[2] Univ Nacl Colombia, Sede Manizales, Dept Informat & Computac, Km 7 Via Magdalena, Manizales 170003, Colombia
[3] Univ Nacl Colombia, Dept Ciencias Computac & Decis, Sede Medellin, Cra 80 65-223, Medellin 050041, Colombia
关键词
Automatic visual inspection; Multi-instance learning; Concept drift; Incremental learning; Non-stationary environments; CLASSIFICATION;
D O I
10.1016/j.compind.2019.04.006
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
Most Multiple Instance Learning (MIL) algorithms are designed with the assumption that the target concept is stationary in time, i.e. it is drawn from a stationary unknown distribution. However, in real industrial applications, like automatic visual inspection where defects may evolve, MIL has to deal with changing target concepts whose statistical characteristics may vary over time. Despite this fact, there is little discussion about how to learn from data in non-stationary environments (or data with concept drift) using multiple instance learners. In this work, an incremental MIL algorithm is proposed in order to learn non-stationary and recurrent target concepts in industrial visual inspection applications. Experiments on both synthetic and real-world datasets are conducted to test the performance of the proposed approach. Real-world datasets come from the automatic visual inspection task in industry. The experimental results show that the proposed approach is able to handle changing target concepts over time. (C) 2019 Elsevier B.V. All rights reserved.
引用
收藏
页码:153 / 164
页数:12
相关论文
共 56 条
[41]   Multiresolution gray-scale and rotation invariant texture classification with local binary patterns [J].
Ojala, T ;
Pietikäinen, M ;
Mäenpää, T .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2002, 24 (07) :971-987
[42]  
Polikar R., 2006, IEEE Circuits and Systems Magazine, V6, P21, DOI 10.1109/MCAS.2006.1688199
[43]   Learn++: An incremental learning algorithm for supervised neural networks [J].
Polikar, R ;
Udpa, L ;
Udpa, SS ;
Honavar, V .
IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART C-APPLICATIONS AND REVIEWS, 2001, 31 (04) :497-508
[44]   Online multiple instance boosting for object detection [J].
Qi, Zhiquan ;
Xu, Yitian ;
Wang, Laisheng ;
Song, Ye .
NEUROCOMPUTING, 2011, 74 (10) :1769-1775
[45]  
Street W. N., 2001, P 7 ACM SIGKDD INT C, P377
[46]   Automatic Metallic Surface Defect Detection and Recognition with Convolutional Neural Networks [J].
Tao, Xian ;
Zhang, Dapeng ;
Ma, Wenzhi ;
Liu, Xilong ;
Xu, De .
APPLIED SCIENCES-BASEL, 2018, 8 (09)
[47]  
Wang J., 2000, P 17 ICML, P1119
[48]  
Wang Shuo, 2010, Neural Networks (IJCNN), The 2010 International Joint Conference on, P1, DOI [10.1109/ISPACS.2010.5704770, DOI 10.1109/IJCNN.2010.5596702]
[49]   Visual tracking with semi-supervised online weighted multiple instance learning [J].
Wang, Zhihui ;
Yoon, Sook ;
Xie, Shan Juan ;
Lu, Yu ;
Park, Dong Sun .
VISUAL COMPUTER, 2016, 32 (03) :307-320
[50]   GPU architecture for unsupervised surface inspection using multi-scale texture analysis [J].
Weimer, Daniel ;
Thamer, Hendrik ;
Thoben, Klaus-Dieter .
2ND INTERNATIONAL CONFERENCE ON SYSTEM-INTEGRATED INTELLIGENCE: CHALLENGES FOR PRODUCT AND PRODUCTION ENGINEERING, 2014, 15 :278-284