Exploiting Counter-Examples for Active Learning with Partial labels
Shanghai Jiao Tong University（上海交通大学）；Hong Kong Baptist University（香港浸会大学）；Nanyang Technological University（南洋理工大学）；Nanjing University of Science and Technology（南京理工大学）
Fei Zhang, Yunjie Ye, Lei Feng, Zhongwen Rao, Jieming Zhu, Marcus Kalander, Chen Gong, Jianye Hao, Bo Han
本文研究了一个新问题——偏标签主动学习（active learning with partial labels，ALPL）。在该设定下，预言机使用偏标签标注查询样本，从而使预言机从要求苛刻的精确标注过程中解脱出来。为解决ALPL，我们首先构建了一个直观的基线，它可以无缝集成到现有的AL框架中。尽管该基线有效，但仍容易受到过拟合的影响，并且在查询过程中缺乏具有代表性的、基于偏标签的样本。受认知科学中人类推理的启发——准确的推理可以明确地从反例（counter-examples，CEs）中得出——我们的目标是利用这种类人学习模式来缓解过拟合，同时增强ALPL中代表性样本的选择过程。具体而言，我们通过反转每个实例的偏标签来构造CEs，随后提出一个简单但有效的WorseNet，直接从这种互补模式中学习。通过利用WorseNet与预测器之间的分布差距，这种对抗性评估方式既可以提升预测器自身的性能，也可以改进样本选择过程，使预测器能够捕捉数据中更准确的模式。在五个真实世界数据集和四个基准数据集上的实验结果表明，我们提出的方法相较十个具有代表性的AL框架取得了全面提升，凸显了WorseNet的优越性。源代码将在 https://github.com/Ferenas/APLL 提供。
英文摘要：
This paper studies a new problem, \emph{active learning with partial labels} (ALPL). In this setting, an oracle annotates the query samples with partial labels, relaxing the oracle from the demanding accurate labeling process. To address ALPL, we first build an intuitive baseline that can be seamlessly incorporated into existing AL frameworks. Though effective, this baseline is still susceptible to the \emph{overfitting}, and falls short of the representative partial-label-based samples during the query process. Drawing inspiration from human inference in cognitive science, where accurate inferences can be explicitly derived from \emph{counter-examples} (CEs), our objective is to leverage this human-like learning pattern to tackle the \emph{overfitting} while enhancing the process of selecting representative samples in ALPL. Specifically, we construct CEs by reversing the partial labels for each instance, and then we propose a simple but effective WorseNet to directly learn from this complementary pattern. By leveraging the distribution gap between WorseNet and the predictor, this adversarial evaluation manner could enhance both the performance of the predictor itself and the sample selection process, allowing the predictor to capture more accurate patterns in the data. Experimental results on five real-world datasets and four benchmark datasets show that our proposed method achieves comprehensive improvements over ten representative AL frameworks, highlighting the superiority of WorseNet. The source code will be available at \url{https://github.com/Ferenas/APLL}.
Exploiting Conjugate Label Information for Multi-Instance Partial-Label Learning
School of Computer Science and Engineering, Southeast University（东南大学计算机科学与工程学院）；Key Lab. of Computer Network and Information Integration (Southeast University), MoE, China（教育部计算机网络与信息集成重点实验室（东南大学））；School of Information and Physical Sciences, The University of Newcastle（纽卡斯尔大学信息与物理科学学院）
Robust Semi-Supervised Learning in Open Environments
Nanjing University（南京大学）；School of Intelligence Science and Technology, Nanjing University（南京大学智能科学与技术学院）；School of Artificial Intelligence, Nanjing University（南京大学人工智能学院）
Counterfactual Reasoning for Multi-Label Image Classification via Patching-Based Training
Nanjing University of Aeronautics and Astronautics（南京航空航天大学）；RIKEN Center for Advanced Intelligence Project（理化学研究所高级智能项目中心）；The University of Tokyo（东京大学）
Exploiting Web Images for Dataset Construction: A Domain Robust Approach
University of Technology Sydney（悉尼科技大学）；University of Electronic Science and Technology of China（电子科技大学）；Alibaba Group（阿里巴巴集团）；Nanjing University of Science and Technology（南京理工大学）
Exploiting Heterogeneous Graph Neural Networks with Latent Worker/Task Correlation Information for Label Aggregation in Crowdsourcing
Zhejiang University（浙江大学）；IBM T. J. Watson Research Center（IBM托马斯·J·沃森研究中心）；JD.COM Silicon Valley Research Center（京东硅谷研发中心）；Squirrel AI Learning（松鼠AI智适应教育）
Exploiting the Potential Supervision Information of Clean Samples in Partial Label Learning
Department of Computer and Information Science, University of Macau（澳门大学计算机与信息科学系）；Department of Computer Science, Hong Kong Baptist University（香港 Baptist 大学计算机科学系）
AllMatch: Exploiting All Unlabeled Data for Semi-Supervised Learning
National Key Laboratory of General Artificial Intelligence（通用人工智能国家重点实验室）；School of Intelligence Science and Technology, Peking University（北京大学智能科学与技术学院）
Exploiting random projections and sparsity with random forests and gradient boosting methods -- Application to multi-label and multi-output learning, random forest model compression and leveraging input sparsity
University of Liège（列日大学）；Aristotle University of Thessaloniki（塞萨洛尼基亚里士多德大学）；Katholieke Universiteit Leuven（鲁汶天主教大学）
Climbing the Kaggle Leaderboard by Exploiting the Log-Loss Oracle
Worcester Polytechnic Institute（伍斯特理工学院）
在数据挖掘竞赛的背景下（例如，Kaggle，KDDCup，ILSVRC挑战），我们展示了如何访问一个甲骨文，报告一个参赛者的日志损失分数的测试集可以被利用来推断地面真理的一些测试的例子。通过迭代地将该技术应用于批量的$m$示例（对于小的$m$），最终可以推断出所有的测试标签。在本文中，（1）我们在最近的Kaggle竞赛（Intel & MobileODT Cancer Screening）的第一阶段展示了这种攻击，并使用它来实现$0.00000$的对数损失（从而在848名参赛者中获得第4名），而无需训练分类器来解决实际任务。(2)我们证明了一个上限的批量大小$m$作为一个函数的浮点分辨率的概率估计，参赛者提交的标签。(3)我们推导出，并在模拟中演示，一个更灵活的攻击，可以使用，即使当预言机报告的准确性测试集的标签的一个未知的（但固定的）子集。这些结果强调了只根据测试数据来评估参赛者的重要性，而甲骨文并不检查这些数据。
ARC-Fi: Exploiting Antenna Spatial Diversity for Label-Efficient Domain Generalization in Wi-Fi Sensing
Suzhou Institute for Advanced Research, University of Science and Technology of China（中国科学技术大学苏州研究院）；Suzhou Big Data and AI Research and Engineering Center（苏州大数据与人工智能研究与工程中心）；School of Artificial Intelligence and Data Science, University of Science and Technology of China（中国科学技术大学人工智能与数据科学学院）；Institute for Infocomm Research (I 2 R), A*STAR（资讯与通讯研究院（I2R），A*STAR）；School of Computer Science and Technology, Soochow University（苏州大学计算机科学与技术学院）
