{"type":"ScholarlyArticle","title":"Positive and Unlabeled Learning via Loss Decomposition and Centroid Estimation","authors":[{"name":"Hong Shi","slug":"hong-shi","given":"Hong","family":"Shi","affiliation":"School of Computer Science and Engineering , Nanjing University of Science and Technology Jiangsu Key Laboratory of Image and Video Understanding for Social Security"},{"name":"Shaojun Pan","slug":"shaojun-pan","given":"Shaojun","family":"Pan","affiliation":"School of Computer Science and Engineering , Nanjing University of Science and Technology Jiangsu Key Laboratory of Image and Video Understanding for Social Security"},{"name":"Jian Yang","slug":"jian-yang","given":"Jian","family":"Yang","affiliation":"School of Computer Science and Engineering , Nanjing University of Science and Technology Jiangsu Key Laboratory of Image and Video Understanding for Social Security"},{"name":"Chen Gong","slug":"chen-gong","given":"Chen","family":"Gong","affiliation":"School of Computer Science and Engineering , Nanjing University of Science and Technology Jiangsu Key Laboratory of Image and Video Understanding for Social Security"}],"abstract":"Positive and Unlabeled learning (PU learning) aims to train a binary classifier based on only positive and unlabeled examples, where the unlabeled examples could be either positive or negative. The state-of-the-art algorithms usually cast PU learning as a cost-sensitive learning problem and impose distinct weights to different training examples via a manual or automatic way. However, such weight adjustment or estimation can be inaccurate and thus often lead to unsatisfactory performance. Therefore, this paper regards all unlabeled examples as negative, which means that some of the original positive data are mistakenly labeled as negative. By doing so, we convert PU learning into the risk minimization problem in the presence of false negative label noise, and propose a novel PU learning algorithm termed ?Loss Decomposition and Centroid Estimation? (LDCE). By decomposing the hinge loss function into two parts, we show that only the second part is influenced by label noise, of which the adverse effect can be reduced by estimating the centroid of negative examples. We intensively validate our approach on synthetic dataset, UCI benchmark datasets and real-world datasets, and the experimental results firmly demonstrate the effectiveness of our approach when compared with other state-of-the-art PU learning methodologies.","doi":"10.24963/ijcai.2018/373","url":"https://ijcai.org/proceedings/2018/373","pdf":"https://ijcai.org/proceedings/2018/0373.pdf","markdown":"https://ijcai.org/proceedings/2018/373.md","bibtex":"https://ijcai.org/proceedings/2018/bibtex/373","volume":{"slug":"2018","title":"Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2018","url":"https://ijcai.org/proceedings/2018/","year":2018,"location":"Stockholm","dates":"13-19 July 2018","editors":"Jérôme Lang","isbn":"978-0-9992411-2-7","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2018"},"section":"Machine Learning","track":"Main track","number":373,"pages":{"first":2689,"last":2695},"keywords":["Machine Learning: Classification","Machine Learning: Machine Learning"],"videos":[],"references":[{"position":1,"raw":"[Cheng et al., 2017] Jiacheng Cheng, Tongliang Liu, Kota- giri Ramamohanarao, and Dacheng Tao. Learning with bounded instance-and label-dependent label noise. arXiv preprint arXiv:1709.03768, 2017.","title":"Learning with bounded instance-and label-dependent label noise","authors":"Cheng","year":"2017","venue":null,"doi":null,"url":null,"internal_url":null},{"position":2,"raw":"[De Comité et al., 1999] Francesco De Comité, Franc ¸ois Denis, Rémi Gilleron, and Fabien Letouzey. Positive and unlabeled examples help learning. In International Con- ference on Algorithmic Learning Theory, pages 219-230, 1999.","title":"Positive and unlabeled examples help learning","authors":"Francesco De Comité, Rémi Franc ¸ois Denis, Fabien Gilleron, Letouzey","year":"1999","venue":"International Conference on Algorithmic Learning Theory","doi":null,"url":null,"internal_url":null},{"position":3,"raw":"[Denis, 1998] Franc ¸ois Denis. PAC learning from positive statistical queries. In International Conference on Algo- rithmic Learning Theory, pages 112-126, 1998.","title":"Franc ¸ois Denis. PAC learning from positive statistical queries","authors":"Denis","year":"1998","venue":"International Conference on Algorithmic Learning Theory","doi":null,"url":null,"internal_url":null},{"position":4,"raw":"[Du Plessis et al., 2014] Marthinus C Du Plessis, Gang Niu, and Masashi Sugiyama. Analysis of learning from positive and unlabeled data. In Advances in Neural Information Processing Systems, pages 703-711, 2014.","title":"Analysis of learning from positive and unlabeled data","authors":"Marthinus C Du, Gang Plessis, Masashi Niu, Sugiyama","year":"2014","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Du Plessis et al., 2015] Marthinus Du Plessis, Gang Niu, and Masashi Sugiyama. Convex formulation for learning from positive and unlabeled data. In International Confer- ence on Machine Learning, pages 1386-1394, 2015.","title":"Convex formulation for learning from positive and unlabeled data","authors":"Marthinus Du Plessis, Gang Niu, Masashi Sugiyama","year":"2015","venue":"International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Elkan and Noto, 2008] Charles Elkan and Keith Noto. Learning classifiers from only positive and unlabeled da- ta. In Proceedings of The 14th ACM SIGKDD Internation- al Conference on Knowledge Discovery and Data Mining, pages 213-220, 2008.","title":"Learning classifiers from only positive and unlabeled data","authors":"Noto Elkan, Charles Elkan, Keith Noto","year":"2008","venue":"Proceedings of The 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Gao et al., 2016] Wei Gao, Lu Wang, Yu-Feng Li, and Zhi- Hua Zhou. Risk minimization in the presence of label noise. In AAAI Conference on Artificial Intelligence, pages 1575-1581, 2016.","title":"Risk minimization in the presence of label noise","authors":"Wei Gao, Lu Wang, Yu-Feng Li, Zhi-Hua Zhou","year":"2016","venue":"AAAI Conference on Artificial Intelligence","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Gong et al., 2015] Chen Gong, Tongliang Liu, Dacheng Tao, Keren Fu, Enmei Tu, and Jie Yang. Deformed graph laplacian for semisupervised learning. IEEE Transaction- s on Neural Networks and Learning Systems, 26:2261- 2274, 2015.","title":"Deformed graph laplacian for semisupervised learning","authors":"Chen Gong, Tongliang Liu, Dacheng Tao, Keren Fu, Enmei Tu, Jie Yang","year":"2015","venue":"IEEE Transactions on Neural Networks and Learning Systems","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Hsu and Sabato, 2014] Daniel Hsu and Sivan Sabato. Heavy-tailed regression with a generalized median-of- means. In International Conference on Machine Learning, pages 37-45, 2014.","title":"Heavy-tailed regression with a generalized median-ofmeans","authors":"Daniel Hsu, Sivan Sabato","year":"2014","venue":"International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Kiryo et al., 2017] Ryuichi Kiryo, Gang Niu, Marthinus C du Plessis, and Masashi Sugiyama. Positive-unlabeled learning with non-negative risk estimator. In Advances in Neural Information Processing Systems, pages 1674- 1684, 2017.","title":"Positive-unlabeled learning with non-negative risk estimator","authors":"Ryuichi Kiryo, Gang Niu, C Marthinus, Masashi Du Plessis, Sugiyama","year":"2017","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Latulippe et al., 2013] Maxime Latulippe, Alexandre Drouin, Philippe Giguere, and Franc ¸ois Laviolette. Accelerated robust point cloud registration in natural environments through positive and unlabeled learning. In International Joint Conference on Artificial Intelligence, pages 2480-2487, 2013.","title":"Accelerated robust point cloud registration in natural environments through positive and unlabeled learning","authors":"Maxime Latulippe, Alexandre Drouin, Philippe Giguere, Franc ¸ois Laviolette","year":"2013","venue":"International Joint Conference on Artificial Intelligence","doi":null,"url":null,"internal_url":"/proceedings/2013/365"},{"position":12,"raw":"[Lee and Liu, 2003] Wee Sun Lee and Bing Liu. Learning with positive and unlabeled examples using weighted lo- gistic regression. In International Conference on Machine Learning, volume 3, pages 448-455, 2003.","title":"Learning with positive and unlabeled examples using weighted logistic regression","authors":"Lee, Liu ; Wee, Sun Lee, Bing Liu","year":"2003","venue":"International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[Li and Liu, 2003] Xiaoli Li and Bing Liu. Learning to clas- sify texts using positive and unlabeled data. In Internation- al Joint Conference on Artificial Intelligence, volume 3, pages 587-592, 2003.","title":"Learning to classify texts using positive and unlabeled data","authors":"Li, Liu ; ] Xiaoli Li, Bing Liu","year":"2003","venue":"International Joint Conference on Artificial Intelligence","doi":null,"url":null,"internal_url":"/proceedings/2003/87"},{"position":14,"raw":"[Liu and Tao, 2016] Tongliang Liu and Dacheng Tao. Clas- sification with noisy labels by importance reweighting. IEEE Transactions on pattern analysis and machine in- telligence, pages 447-461, 2016.","title":"Tongliang Liu and Dacheng Tao. Classification with noisy labels by importance reweighting","authors":"Tao Liu","year":"2016","venue":"IEEE Transactions on pattern analysis and machine intelligence","doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Liu et al., 2002] Bing Liu, Wee Sun Lee, Philip S Yu, and Xiaoli Li. Partially supervised classification of text docu- ments. In International Conference on Machine Learning, pages 387-394, 2002.","title":"Partially supervised classification of text documents","authors":"Bing Liu, Wee Sun Lee, Philip S Yu, Xiaoli Li","year":"2002","venue":"International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Liu et al., 2003] Bing Liu, Yang Dai, Xiaoli Li, Wee Sun Lee, and Philip S Yu. Building text classifiers using pos- itive and unlabeled examples. In The IEEE International Conference on Data Mining, volume 2, pages 179-186, 2003.","title":"Building text classifiers using positive and unlabeled examples","authors":"Bing Liu, Yang Dai, Xiaoli Li, Wee Sun Lee, Philip S Yu","year":"2003","venue":"The IEEE International Conference on Data Mining","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Merz and Murphy, 1998] Christopher J Merz and Patrick M Murphy. UCI Repository of machine learning databases. 1998.","title":"UCI Repository of machine learning databases","authors":"J Christopher, Patrick M Merz, Murphy","year":"1998","venue":null,"doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Natarajan et al., 2013] Nagarajan Natarajan, Inderjit S D- hillon, Pradeep K Ravikumar, and Ambuj Tewari. Learn- ing with noisy labels. In Advances in Neural Information Processing Systems, pages 1196-1204, 2013.","title":"Learning with noisy labels","authors":"Nagarajan Natarajan, Pradeep K Inderjit S Dhillon, Ambuj Ravikumar, Tewari","year":"2013","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Nigam et al., 1998] Kamal Nigam, Andrew McCallum, Se- bastian Thrun, Tom Mitchell, et al. Learning to classify text from labeled and unlabeled documents. In AAAI Con- ference on Artificial Intelligence, volume 792, 1998.","title":"Learning to classify text from labeled and unlabeled documents","authors":"Kamal Nigam, Andrew Mccallum, Sebastian Thrun, Tom Mitchell","year":"1998","venue":"AAAI Conference on Artificial Intelligence","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Patrini et al., 2016] Giorgio Patrini, Frank Nielsen, Richard Nock, and Marcello Carioni. Loss factorization, weakly supervised learning and label noise robustness. In Interna- tional Conference on Machine Learning, pages 708-717, 2016.","title":"Loss factorization, weakly supervised learning and label noise robustness","authors":"Giorgio Patrini, Frank Nielsen, Richard Nock, Marcello Carioni","year":"2016","venue":"International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Scott and Blanchard, 2009] Clayton Scott and Gilles Blan- chard. Novelty detection: Unlabeled data definitely help. In Artificial Intelligence and Statistics, pages 464-471, 2009.","title":"Novelty detection: Unlabeled data definitely help","authors":"Clayton Scott, Gilles Blanchard","year":"2009","venue":"Artificial Intelligence and Statistics","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Yu et al., 2017a] Xiyu Yu, Tongliang Liu, Mingming Gong, and Dacheng Tao. Learning with biased complementary labels. arXiv preprint arXiv:1711.09535, 2017.","title":"Learning with biased complementary labels","authors":"Yu","year":"2017","venue":null,"doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Yu et al., 2017b] Xiyu Yu, Tongliang Liu, Mingming Gong, Kun Zhang, and Dacheng Tao. Transfer learning with label noise. arXiv preprint arXiv:1707.09724, 2017.","title":"Transfer learning with label noise","authors":"Yu","year":"2017","venue":null,"doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}