{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,27]],"date-time":"2026-05-27T22:24:38Z","timestamp":1779920678021,"version":"3.53.1"},"reference-count":35,"publisher":"MIT Press","issue":"1","content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2023,1,1]]},"abstract":"<jats:title>Abstract<\/jats:title><jats:p>Partial-label learning is a kind of weakly supervised learning with inexact labels, where for each training example, we are given a set of candidate labels instead of only one true label. Recently, various approaches on partial-label learning have been proposed under different generation models of candidate label sets. However, these methods require relatively strong distributional assumptions on the generation models. When the assumptions do not hold, the performance of the methods is not guaranteed theoretically. In this letter, we propose the notion of properness on partial labels. We show that this proper partial-label learning framework requires a weaker distributional assumption and includes many previous partial-label learning settings as special cases. We then derive a unified unbiased estimator of the classification risk. We prove that our estimator is risk consistent, and we also establish an estimation error bound. Finally, we validate the effectiveness of our algorithm through experiments.<\/jats:p>","DOI":"10.1162\/neco_a_01554","type":"journal-article","created":{"date-parts":[[2022,11,24]],"date-time":"2022-11-24T13:59:53Z","timestamp":1669298393000},"page":"58-81","update-policy":"https:\/\/2.zoppoz.workers.dev:443\/https\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":5,"title":["Learning With Proper Partial Labels"],"prefix":"10.1162","volume":"35","author":[{"given":"Zhenguo","family":"Wu","sequence":"first","affiliation":[{"name":"University of Tokyo, Bunkyo, Tokyo 113-0033, Japan zhenguo@ms.k.u-tokyo.ac.jp"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jiaqi","family":"Lv","sequence":"additional","affiliation":[{"name":"RIKEN AIP, Tokyo 103-0027, Japan jiaqi.lyu@riken.jp"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Masashi","family":"Sugiyama","sequence":"additional","affiliation":[{"name":"RIKEN AIP, Tokyo 103-0027, Japan"},{"name":"University of Tokyo, Bunkyo, Tokyo 113-0033, Japan sugi@k.u-tokyo.ac.jp"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"281","published-online":{"date-parts":[[2023,1,1]]},"reference":[{"key":"2023032023361013900_","first-page":"452","article-title":"Classification from pairwise similarity and unlabeled data","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Bao","year":"2018"},{"key":"2023032023361013900_","first-page":"825","article-title":"Confidence scores make instance-dependent label-noise learning possible","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Berthon","year":"2021"},{"key":"2023032023361013900_","author":"Cao","year":"2021","journal-title":"Multi-class classification from single-class data with confidences"},{"key":"2023032023361013900_","first-page":"1272","article-title":"Learning from similarity-confidence data","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Cao","year":"2021"},{"key":"2023032023361013900_","first-page":"961","article-title":"On symmetric losses for learning from corrupted labels","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Charoenphakdee","year":"2019"},{"key":"2023032023361013900_","author":"Clanuwat","year":"2018","journal-title":"Deep learning for classical Japanese literature"},{"key":"2023032023361013900_","first-page":"1501","article-title":"Learning from partial labels","volume":"12","author":"Cour","year":"2011","journal-title":"Journal of Machine Learning Research"},{"key":"2023032023361013900_","first-page":"3072","article-title":"Learning with multiple complementary labels","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Feng","year":"2020"},{"key":"2023032023361013900_","first-page":"10948","article-title":"Provably consistent partial-label learning","volume-title":"Advances in neural information processing systems","author":"Feng","year":"2020"},{"key":"2023032023361013900_","first-page":"3252","article-title":"Pointwise binary classification with pairwise confidence comparisons","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Feng","year":"2021"},{"key":"2023032023361013900_","first-page":"5836","article-title":"Masking: A new perspective of noisy supervision","volume-title":"Advances in neural information processing systems","author":"Han","year":"2018"},{"key":"2023032023361013900_","first-page":"770","article-title":"Deep residual learning for image recognition","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"He","year":"2016"},{"key":"2023032023361013900_","first-page":"4700","article-title":"Densely connected convolutional networks","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Huang","year":"2017"},{"key":"2023032023361013900_","first-page":"448","article-title":"Batch normalization: Accelerating deep network training by reducing internal covariate shift","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Ioffe","year":"2015"},{"key":"2023032023361013900_","first-page":"5639","article-title":"Learning from complementary labels","volume-title":"Advances in neural information processing systems","author":"Ishida","year":"2017"},{"key":"2023032023361013900_","first-page":"2971","article-title":"Complementary-label learning for arbitrary losses and models","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Ishida","year":"2019"},{"key":"2023032023361013900_","first-page":"5917","article-title":"Binary classification from positive-confidence data","volume-title":"Advances in neural information processing systems","author":"Ishida","year":"2018"},{"key":"2023032023361013900_","volume-title":"Learning multiple layers of features from tiny images","author":"Krizhevsky","year":"2009"},{"key":"2023032023361013900_","first-page":"2278","article-title":"Gradient-based learning applied to document recognition","volume-title":"Proceedings of the IEEE","author":"LeCun","year":"1998"},{"key":"2023032023361013900_","first-page":"548","article-title":"A conditional multinomial mixture model for superset label learning","volume-title":"Advances in neural information processing systems","author":"Liu","year":"2012"},{"key":"2023032023361013900_","first-page":"1629","article-title":"Learnability of the superset label learning problem","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Liu","year":"2014"},{"key":"2023032023361013900_","article-title":"On the minimal supervision for training any binary classifier from only unlabeled data","volume-title":"Proceedings of the International Conference on Learning Representations","author":"Lu","year":"2019"},{"key":"2023032023361013900_","first-page":"1504","article-title":"Learning from candidate labeling sets","volume-title":"Advances in neural information processing systems","author":"Luo","year":"2010"},{"key":"2023032023361013900_","first-page":"6500","article-title":"Progressive identification of true labels for partial-label learning","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Lv","year":"2020"},{"key":"2023032023361013900_","doi-asserted-by":"crossref","first-page":"3","DOI":"10.1007\/978-3-319-46379-7_1","article-title":"A vector-contraction inequality for Rademacher complexities","volume-title":"Proceedings of the International Conference on Algorithmic Learning Theory","author":"Maurer","year":"2016"},{"issue":"1","key":"2023032023361013900_","first-page":"148","article-title":"On the method of bounded differences","volume":"141","author":"McDiarmid","year":"1989","journal-title":"Surveys in Combinatorics"},{"key":"2023032023361013900_","volume-title":"Foundations of machine learning","author":"Mohri","year":"2018"},{"issue":"3","key":"2023032023361013900_","doi-asserted-by":"publisher","first-page":"400","DOI":"10.1214\/aoms\/1177729586","article-title":"A stochastic approximation method","volume":"22","author":"Robbins","year":"1951","journal-title":"Annals of Mathematical Statistics"},{"key":"2023032023361013900_","volume-title":"Machine learning from weak supervision: An empirical risk minimization approach","author":"Sugiyama","year":"2022"},{"key":"2023032023361013900_","first-page":"11091","article-title":"Leveraged weighted loss for partial label learning","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Wen","year":"2021"},{"key":"2023032023361013900_","author":"Xiao","year":"2017","journal-title":"Fashion-MNIST: A novel image dataset for benchmarking machine learning algorithms"},{"key":"2023032023361013900_","first-page":"68","article-title":"Learning with biased complementary labels","volume-title":"Proceedings of the European Conference on Computer Vision","author":"Yu","year":"2018"},{"key":"2023032023361013900_","first-page":"708","article-title":"Learning by associating ambiguously labeled images","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Zeng","year":"2013"},{"key":"2023032023361013900_","first-page":"4048","article-title":"Solving the partial label learning problem: An instance-based approach","volume-title":"Proceedings of the International Joint Conference on Artificial Intelligence","author":"Zhang","year":"2015"},{"issue":"1","key":"2023032023361013900_","doi-asserted-by":"publisher","first-page":"44","DOI":"10.1093\/nsr\/nwx106","article-title":"A brief introduction to weakly supervised learning","volume":"5","author":"Zhou","year":"2018","journal-title":"National Science Review"}],"container-title":["Neural Computation"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/direct.mit.edu\/neco\/article-pdf\/35\/1\/58\/2075432\/neco_a_01554.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/direct.mit.edu\/neco\/article-pdf\/35\/1\/58\/2075432\/neco_a_01554.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,12,1]],"date-time":"2023-12-01T17:32:40Z","timestamp":1701451960000},"score":1,"resource":{"primary":{"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/direct.mit.edu\/neco\/article\/35\/1\/58\/113808\/Learning-With-Proper-Partial-Labels"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,1,1]]},"references-count":35,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2023,1,1]]},"published-print":{"date-parts":[[2023,1,1]]}},"URL":"https:\/\/2.zoppoz.workers.dev:443\/https\/doi.org\/10.1162\/neco_a_01554","relation":{},"ISSN":["0899-7667","1530-888X"],"issn-type":[{"value":"0899-7667","type":"print"},{"value":"1530-888X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2023,1]]},"published":{"date-parts":[[2023,1,1]]}}}