Self-supervised learning (SSL) is an efficient approach that addresses the issue of limited training data and annotation shortage. The key part in SSL is its proxy task that defines the supervisory signals and drives the learning toward effective feature representations. However, most SSL approaches usually focus on a single proxy task, which greatly limits the expressive power of the learned features and therefore deteriorates the network generalization capacity. In this regard, we hereby propose two strategies of aggregation in terms of complementarity of various forms to boost the robustness of self-supervised learned features. We firstly propose a principled framework of multi-task aggregative self-supervised learning from a limited sample to form a unified representation, with an intent of exploiting feature complementarity among different tasks. Then, in self-aggregative SSL, we propose to self-complement an existing proxy task with an auxiliary loss function based on a linear centered kernel alignment metric, which explicitly promotes the exploring of where are uncovered by the features learned from a proxy task at hand to further boost the modeling capability. Our extensive experiments on 2D natural image and 3D medical image classification tasks under limited data and annotation scenarios confirm that the proposed aggregation strategies successfully boost the classification accuracy.
Semi-Supervised Disease Classification based on Limited Medical Image Data
School of Science, Minzu University of China(中央民族大学理学院);Joint Research Center on Computational Mathematics and Control, Shenzhen MSU-BIT University(深圳北理莫斯科大学计算数学与控制联合研究中心);Institute of Data Science, National University of Singapore(新加坡国立大学数据科学研究院)
Max-Planck Institute of Molecular Cell Biology and Genetics(马克斯·普朗克分子细胞生物学与遗传学研究所);Center for Systems Biology Dresden (CSBD)(德累斯顿系统生物学中心);Fondazione Human Technopole(人类技术城基金会);Max Planck Institute for the Physics of Complex Systems(马克斯·普朗克复杂系统物理研究所)
Model-based Reconstruction with Learning: From Unsupervised to Supervised and Beyond
Michigan State University(密歇根州立大学);University of Michigan - Shanghai Jiao Tong University Joint Institute(密歇根大学-上海交通大学联合学院);Shanghai Jiao Tong University(上海交通大学)
Semi- and Weakly-Supervised Learning for Mammogram Mass Segmentation with Limited Annotations
School of Computer Science and Engineering, Sun Yat-sen University(中山大学计算机科学与工程学院);School of Computer Science, Peking University(北京大学计算机科学学院);School of Future Technology, Tianjin University(天津大学未来技术学院);Research Institute, Sun Yat-sen University(中山大学研究学院)
Learning with Limited Annotations: A Survey on Deep Semi-Supervised Learning for Medical Image Segmentation
School of Biomedical Engineering, Shanghai Jiao Tong University(上海交通大学生物医学工程学院);School of Data Science, Fudan University(复旦大学大数据学院);School of Engineering Medicine, Beihang University(北京航空航天大学医工交叉创新研究院);Artificial Intelligence Innovation and Incubation Institute, Fudan University(复旦大学人工智能创新与产业研究院);School of Biological Science and Medical Engineering, Beihang University(北京航空航天大学生物与医学工程学院);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);Beijing Anding Hospital, Capital Medical University(首都医科大学附属北京安定医院);Key Laboratory for Biomechanics and Mechanobiology of Ministry of Education, Beihang University(北京航空航天大学生物力学与力学生物学教育部重点实验室);Hefei Innovation Research Institute, Beihang University(北京航空航天大学合肥创新研究院)
Active Self-Semi-Supervised Learning for Few Labeled Samples
Australian Centre for Robotics, The University of Sydney(悉尼大学澳大利亚机器人中心);Department of Marine Technology, Norwegian University of Science and Technology(挪威科技大学海洋技术系)
Self-aware and Cross-sample Prototypical Learning for Semi-supervised Medical Image Segmentation
Xidian University(西安电子科技大学);The First Affiliated Hospital of Xi’an Jiaotong University(西安交通大学第一附属医院);AIQ;Warren Alpert Medical School of Brown University(布朗大学沃伦·阿尔珀特医学院)
Inter- and intra-uncertainty based feature aggregation model for semi-supervised histopathology image segmentation
Northwestern Polytechnical University(西北工业大学);La Trobe University(拉筹伯大学);CSIRO Data61(联邦科学与工业研究组织数据61部);University of New South Wales(新南威尔士大学);Shandong University(山东大学);Tianjin University(天津大学)
Self-Supervised Learning with Masked Image Modeling for Teeth Numbering, Detection of Dental Restorations, and Instance Segmentation in Dental Panoramic Radiographs
X-ModalNet: A Semi-Supervised Deep Cross-Modal Network for Classification of Remote Sensing Data
German Aerospace Center(德国航空航天中心);Technical University of Munich(慕尼黑工业大学);The University of Tokyo(东京大学);RIKEN(日本理化学研究所);Wuhan University(武汉大学);Univ. Grenoble Alpes(格勒诺布尔阿尔卑斯大学);INRIA(法国国家信息与自动化研究所);CNRS(法国国家科学研究中心);Grenoble INP(格勒诺布尔理工学院);Aerospace Information Research Institute, Chinese Academy of Sciences(中国科学院空天信息创新研究院)
Leveraging Semi-Supervised Learning to Enhance Data Mining for Image Classification under Limited Labeled Data
University of Michigan(密歇根大学);Columbia University(哥伦比亚大学);Tulane University(杜兰大学);Northeastern University(东北大学);Dartmouth College(达特茅斯学院);Shanghai Jiao Tong University(上海交通大学)
Uncertainty-Aware Semi-Supervised Method Using Large Unlabeled and Limited Labeled COVID-19 Data
Deakin University(迪肯大学);Isfahan University of Technology(伊斯法罕理工大学);University of Birjand(比尔詹德大学);Ferdowsi University of Mashhad(马什哈德菲尔多西大学);K. N. Toosi University of Technology(图西理工大学);Universidad de Granada(格拉纳达大学);Dibrugarh University(迪布鲁格尔大学);Iran University of Medical Sciences(伊朗医科大学);The George Institute for Global Health(乔治全球健康研究院);University of Sydney(悉尼大学);Ngee Ann Polytechnic(义安理工学院);Singapore University of Social Sciences(新加坡新跃社科大学);Asia University(亚洲大学)
FATE: A Prompt-Tuning-Based Semi-Supervised Learning Framework for Extremely Limited Labeled Data
Xiamen University(厦门大学);Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ)(广东省人工智能与数字经济实验室(深圳));Guangdong University of Technology(广东工业大学);University of Nottingham(诺丁汉大学);Nanjing University of Science and Technology(南京理工大学)
A Semi-Supervised Framework for Breast Ultrasound Segmentation with Training-Free Pseudo-Label Generation and Label Refinement
Tohoku University Graduate School of Medicine(东北大学医学研究科);Advanced Institute of Convergence Knowledge Informatics(融合知识信息先进研究所);Research Institute of Electrical Communication, Tohoku University(东北大学电气通信研究所);National Institute of Technology, Sendai College(名古屋国立技术大学送崎学院);State Key Laboratory of Oncology in South China, Sun Yat-sen University Cancer Center(南方肿瘤学国家重点实验室,中山大学肿瘤中心);Department of Mathematics, Informatics, and Cybernetics, University of Chemistry and Technology(化学与技术大学数学、信息学与自动控制系);School of Software Technology, Zhejiang University(浙江大学软件技术学院);Southeast University, School of Cyber Science and Engineering(东南大学网络科学与工程学院)