{"@context":{"@vocab":"https://cir.nii.ac.jp/schema/1.0/","rdfs":"http://www.w3.org/2000/01/rdf-schema#","dc":"http://purl.org/dc/elements/1.1/","dcterms":"http://purl.org/dc/terms/","foaf":"http://xmlns.com/foaf/0.1/","prism":"http://prismstandard.org/namespaces/basic/2.0/","cinii":"http://ci.nii.ac.jp/ns/1.0/","datacite":"https://schema.datacite.org/meta/kernel-4/","ndl":"http://ndl.go.jp/dcndl/terms/","jpcoar":"https://github.com/JPCOAR/schema/blob/master/2.0/"},"@id":"https://cir.nii.ac.jp/crid/1360021390582777344.json","@type":"Article","productIdentifier":[{"identifier":{"@type":"DOI","@value":"10.1109/iccv51070.2023.01519"}},{"identifier":{"@type":"URI","@value":"http://xplorestaging.ieee.org/ielx7/10376473/10376477/10378601.pdf?arnumber=10378601"}},{"identifier":{"@type":"DOI","@value":"10.48550/arxiv.2308.08822"}}],"resourceType":"学術雑誌論文(journal article)","dc:title":[{"@value":"MixBag: Bag-Level Data Augmentation for Learning from Label Proportions"}],"description":[{"notation":[{"@value":"Learning from label proportions (LLP) is a promising weakly supervised learning problem. In LLP, a set of instances (bag) has label proportions, but no instance-level labels are given. LLP aims to train an instance-level classifier by using the label proportions of the bag. In this paper, we propose a bag-level data augmentation method for LLP called MixBag, based on the key observation from our preliminary experiments; that the instance-level classification accuracy improves as the number of labeled bags increases even though the total number of instances is fixed. We also propose a confidence interval loss designed based on statistical theory to use the augmented bags effectively. To the best of our knowledge, this is the first attempt to propose bag-level data augmentation for LLP. The advantage of MixBag is that it can be applied to instance-level data augmentation techniques and any LLP method that uses the proportion loss. Experimental results demonstrate this advantage and the effectiveness of our method."}]},{"notation":[{"@value":"Accepted at ICCV2023"}]}],"creator":[{"@id":"https://cir.nii.ac.jp/crid/1380021390582777482","@type":"Researcher","foaf:name":[{"@value":"Takanori Asanomi"}],"jpcoar:affiliationName":[{"@value":"Kyushu University,Fukuoka,Japan"}]},{"@id":"https://cir.nii.ac.jp/crid/1380021390582777475","@type":"Researcher","foaf:name":[{"@value":"Shinnosuke Matsuo"}],"jpcoar:affiliationName":[{"@value":"Kyushu University,Fukuoka,Japan"}]},{"@id":"https://cir.nii.ac.jp/crid/1380021390582777485","@type":"Researcher","foaf:name":[{"@value":"Daiki Suehiro"}],"jpcoar:affiliationName":[{"@value":"Kyushu University,Fukuoka,Japan"}]},{"@id":"https://cir.nii.ac.jp/crid/1420564276158799744","@type":"Researcher","personIdentifier":[{"@type":"KAKEN_RESEARCHERS","@value":"00644270"},{"@type":"NRID","@value":"1000000644270"},{"@type":"NRID","@value":"9000004750874"},{"@type":"NRID","@value":"9000411115720"},{"@type":"NRID","@value":"9000414365777"},{"@type":"RESEARCHMAP","@value":"https://researchmap.jp/biser"}],"foaf:name":[{"@value":"Ryoma Bise"}],"jpcoar:affiliationName":[{"@value":"Kyushu University,Fukuoka,Japan"}]}],"publication":{"prism:publicationName":[{"@value":"2023 IEEE/CVF International Conference on Computer Vision (ICCV)"}],"dc:publisher":[{"@value":"IEEE"}],"prism:publicationDate":"2023-10-01","prism:startingPage":"16524","prism:endingPage":"16533"},"reviewed":"false","dcterms:accessRights":"http://purl.org/coar/access_right/c_abf2","dc:rights":["https://doi.org/10.15223/policy-029","https://doi.org/10.15223/policy-037"],"url":[{"@id":"http://xplorestaging.ieee.org/ielx7/10376473/10376477/10378601.pdf?arnumber=10378601"}],"createdAt":"2024-01-15","modifiedAt":"2024-01-18","foaf:topic":[{"@id":"https://cir.nii.ac.jp/all?q=FOS:%20Computer%20and%20information%20sciences","dc:title":"FOS: Computer and information sciences"},{"@id":"https://cir.nii.ac.jp/all?q=Computer%20Vision%20and%20Pattern%20Recognition%20(cs.CV)","dc:title":"Computer Vision and Pattern Recognition (cs.CV)"},{"@id":"https://cir.nii.ac.jp/all?q=Computer%20Science%20-%20Computer%20Vision%20and%20Pattern%20Recognition","dc:title":"Computer Science - Computer Vision and Pattern Recognition"}],"project":[{"@id":"https://cir.nii.ac.jp/crid/1040296807938645760","@type":"Project","projectIdentifier":[{"@type":"KAKEN","@value":"23K18509"},{"@type":"JGN","@value":"JP23K18509"},{"@type":"URI","@value":"https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-23K18509/"}],"notation":[{"@language":"ja","@value":"病理診断での実応用のための部分クラス比率学習を用いた弱教師あり学習手法の開発"},{"@language":"en","@value":"Learning from partial label proportion for pathological image diagnosis"}]},{"@id":"https://cir.nii.ac.jp/crid/1040858906220008064","@type":"Project","projectIdentifier":[{"@type":"KAKEN","@value":"23KJ1723"},{"@type":"JGN","@value":"JP23KJ1723"},{"@type":"URI","@value":"https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-23KJ1723/"}],"notation":[{"@language":"ja","@value":"組み合わせ最適化問題への画像解析的アプローチ"}]}],"relatedProduct":[{"@id":"https://cir.nii.ac.jp/crid/1050289321218079744","@type":"Article","resourceType":"会議発表資料(conference paper)","relationType":["references"],"jpcoar:relatedTitle":[{"@language":"en","@value":"Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology"}]},{"@id":"https://cir.nii.ac.jp/crid/1360016866527693440","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Empirical Risk Minimization"}]},{"@id":"https://cir.nii.ac.jp/crid/1360294646123847936","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"A novel multiple instance learning framework for COVID-19 severity assessment via data augmentation and self-supervised learning"}]},{"@id":"https://cir.nii.ac.jp/crid/1360302866849844480","@type":"Article","resourceType":"学術雑誌論文(journal article)","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Learning From Label Proportion with Online Pseudo-Label Decision by Regret Minimization"}]},{"@id":"https://cir.nii.ac.jp/crid/1360302869453669248","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Co-Training for Demographic Classification Using Deep Learning from Label Proportions"}]},{"@id":"https://cir.nii.ac.jp/crid/1360302870939888000","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"ReMix: A General and Efficient Framework for Multiple Instance Learning Based Whole Slide Image Classification"}]},{"@id":"https://cir.nii.ac.jp/crid/1360306904393420032","@type":"Article","resourceType":"学術雑誌論文(journal article)","relationType":["isReferencedBy"],"jpcoar:relatedTitle":[{"@value":"Rethinking Multiple Instance Learning for Whole Slide Image Classification: A Bag-Level Classifier is a Good Instance-Level Teacher"}]},{"@id":"https://cir.nii.ac.jp/crid/1360579817926265728","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"The MNIST Database of Handwritten Digit Images for Machine Learning Research [Best of the Web]"}]},{"@id":"https://cir.nii.ac.jp/crid/1360584346107729792","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Two-stage Training for Learning from Label Proportions"}]},{"@id":"https://cir.nii.ac.jp/crid/1360855567865740544","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Deep Residual Learning for Image Recognition"}]},{"@id":"https://cir.nii.ac.jp/crid/1360865818730350848","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"MedMNIST v2 - A large-scale lightweight benchmark for 2D and 3D biomedical image classification"}]},{"@id":"https://cir.nii.ac.jp/crid/1360865820416596992","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Unified Generative Adversarial Networks for Controllable Image-to-Image Translation"}]},{"@id":"https://cir.nii.ac.jp/crid/1360869854351437312","@type":"Article","resourceType":"学術雑誌論文(journal article)","relationType":["isReferencedBy"],"jpcoar:relatedTitle":[{"@value":"Learning from Partial Label Proportions for Whole Slide Image Segmentation"}]},{"@id":"https://cir.nii.ac.jp/crid/1361137044582314752","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Randaugment: Practical automated data augmentation with a reduced search space"}]},{"@id":"https://cir.nii.ac.jp/crid/1361699994757714944","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features"}]},{"@id":"https://cir.nii.ac.jp/crid/1361981469930876288","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Learning With Label Proportions via NPSVM"}]},{"@id":"https://cir.nii.ac.jp/crid/1361981471340537600","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"AutoAugment: Learning Augmentation Strategies From Data"}]},{"@id":"https://cir.nii.ac.jp/crid/1363670319058859264","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"A survey on Image Data Augmentation for Deep Learning"}]},{"@id":"https://cir.nii.ac.jp/crid/1363670319352108544","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Principal component analysis"}]},{"@id":"https://cir.nii.ac.jp/crid/1363670320867785088","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"ImageNet: A large-scale hierarchical image database"}]},{"@id":"https://cir.nii.ac.jp/crid/1363951793834076288","@type":"Article","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777345","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Improved regularization of convolutional neural networks with cutout"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777346","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Pytorch: An imperative style, high-performance deep learning library"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777348","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"A two-stage training framework with feature-label matching mechanism for learning from label proportions"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777472","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Fast autoaugment"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777474","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Fencemask: a data augmentation approach for pre-extracted image features"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777479","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Deep multi-class learning from label proportions"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777480","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Temporal ensembling for semisupervised learning"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777483","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Learning from label proportions with consistency regularization"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777485","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Svm classifier estimation from group probabilities"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777489","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Learning from label proportions with generative adversarial networks"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777496","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Augmix: A simple data processing method to improve robustness and uncertainty"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777497","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Data augmentation by pairing samples for images classification"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777498","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"Gridmask data augmentation"}]},{"@id":"https://cir.nii.ac.jp/crid/1370021390582777499","@type":"Product","relationType":["references"],"jpcoar:relatedTitle":[{"@value":"The effectiveness of data augmentation in image classification using deep learning"}]}],"dataSourceIdentifier":[{"@type":"CROSSREF","@value":"10.1109/iccv51070.2023.01519"},{"@type":"KAKEN","@value":"PRODUCT-24839885"},{"@type":"KAKEN","@value":"PRODUCT-24783456"},{"@type":"OPENAIRE","@value":"doi_dedup___::0f6b212e4a6bac5633ce27730566d7fa"},{"@type":"CROSSREF","@value":"10.1109/tmi.2024.3404549_references_DOI_Qi6eU8CBxrCNr5oTpahVrNtofW4"},{"@type":"CROSSREF","@value":"10.1007/978-3-031-72120-5_35_references_DOI_Qi6eU8CBxrCNr5oTpahVrNtofW4"}]}