Multi-view learning attempts to generate a model with a better performance by exploiting the consensus and/or complementarity among multi-view data. However, in terms of complementarity, most existing approaches only can find representations with single complementarity rather than complementary information with diversity. In this paper, to utilize both complementarity and consistency simultaneously, give free rein to the potential of deep learning in grasping diversity-promoting complementarity for multi-view representation learning, we propose a novel supervised multi-view representation learning algorithm, called Self-Attention Multi-View network with Diversity-Promoting Complementarity (SAMVDPC), which exploits the consistency by a group of encoders, uses self-attention to find complementary information entailing diversity. Extensive experiments conducted on eight real-world datasets have demonstrated the effectiveness of our proposed method, and show its superiority over several baseline methods, which only consider single complementary information.
DiRe: Diversity-promoting Regularization for Dataset Condensation
Department of Artificial Intelligence, Indian Institute of Technology Hyderabad(人工智能系,印度理工学院海得拉巴);Centre for Responsible AI, Wadhwani School of Data Science & AI, Indian Institute of Technology Madras(负责任人工智能中心,Wadhwani数据科学与人工智能学院,印度理工学院马德拉斯)
The Minority Matters: A Diversity-Promoting Collaborative Metric Learning Algorithm
Institute of Information Engineering, CAS(中国科学院信息工程研究所);University of Chinese Academy of Sciences(中国科学院大学);Institute of Computing Technology, CAS(中国科学院计算技术研究所);Alibaba Group(阿里巴巴集团);Sun Yat-sen University(中山大学);Peng Cheng Laboratory(鹏城实验室)
协作度量学习(CML)最近成为推荐系统(RS)中的一种流行方法,缩小了度量学习和协作过滤之间的差距。遵循RS的惯例,现有的方法在其模型设计中利用唯一的用户表示。本文重点介绍了一个具有挑战性的情况下,用户有多个类别的利益。在这种情况下,我们认为,独特的用户表示可能会导致偏好偏差,特别是当项目类别分布不平衡。为了解决这个问题,我们提出了一种新的方法称为\textit{Diversity-Promoting Collaborative Metric Learning}(DPCML),希望考虑到通常被忽视的少数用户的利益。DPCML背后的关键思想是为系统中的每个用户包括多个表示集。基于此嵌入范式,用户对一个项目的偏好聚合从不同的嵌入之间的最小项目用户的距离用户嵌入集。此外,我们观察到,同一用户的嵌入的多样性在模型中也起着至关重要的作用。为此,我们提出了一个\textit{diversity control regularization}术语,以更好地适应多向量表示策略。从理论上讲,我们表明,DPCML可以很好地推广到看不见的测试数据,通过解决来自最小值的恼人的操作的挑战。在一系列基准数据集上的实验证明了DPCML的有效性。
Improved Diversity-Promoting Collaborative Metric Learning for Recommendation
Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所);University of Chinese Academy of Sciences(中国科学院大学);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);Alibaba Group(阿里巴巴集团);Sun Yat-sen University(中山大学);Peng Cheng Laboratory(鹏城实验室)
Promoting Ensemble Diversity with Interactive Bayesian Distributional Robustness for Fine-tuning Foundation Models
Qualcomm AI Research, Qualcomm Vietnam Company Limited(高通人工智能研究, 高通越南公司);National University of Singapore, Singapore(新加坡国立大学);Monash University, Australia(墨尔本大学)
Evaluation of Self-taught Learning-based Representations for Facial Emotion Recognition
Pontifícia Universidade Católica do Paraná (PUCPR)(巴拉那天主教大学(PUCPR));State University of Ponta Grossa (UEPG)(蓬塔格罗萨州立大学(UEPG));École de Technologie Supérieure (ÉTS), Université du Québec(魁北克大学高等技术学院(ÉTS))
Contrastive Representation Modeling for Anomaly Detection
Technology Innovation Institute, Abu Dhabi, UAE(阿布扎比技术创新研究所);Computer Science Department, Khalifa University of Science and Technology, Abu Dhabi, UAE(科技学院计算机科学系)
Mixture of Complementary Agents for Robust LLM Ensemble
DIMACS, Rutgers University(罗格斯大学DIMACS研究中心);Department of Mathematics, Rutgers University(罗格斯大学数学系);Department of Computer Science, George Mason University(乔治·梅森大学计算机科学系);Department of Computer Science, Rutgers University(罗格斯大学计算机科学系)