Enhanced Multimodal Representation Learning with Cross-modal KD

被引:3
|
作者
Chen, Mengxi [1 ]
Xing, Linyu [1 ]
Wang, Yu [1 ,2 ]
Zhang, Ya [1 ,2 ]
机构
[1] Shanghai Jiao Tong Univ, Shanghai, Peoples R China
[2] Shanghai AI Lab, Shanghai, Peoples R China
基金
国家重点研发计划;
关键词
NETWORKS;
D O I
10.1109/CVPR52729.2023.01132
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper explores the tasks of leveraging auxiliary modalities which are only available at training to enhance multimodal representation learning through cross-modal Knowledge Distillation (KD). The widely adopted mutual information maximization-based objective leads to a short-cut solution of the weak teacher, i.e., achieving the maximum mutual information by simply making the teacher model as weak as the student model. To prevent such a weak solution, we introduce an additional objective term, i.e., the mutual information between the teacher and the auxiliary modality model. Besides, to narrow down the information gap between the student and teacher, we further propose to minimize the conditional entropy of the teacher given the student. Novel training schemes based on contrastive learning and adversarial learning are designed to optimize the mutual information and the conditional entropy, respectively. Experimental results on three popular multimodal benchmark datasets have shown that the proposed method outperforms a range of state-of-the-art approaches for video recognition, video retrieval and emotion classification.
引用
收藏
页码:11766 / 11775
页数:10
相关论文
共 50 条
  • [1] Multimodal Reaction: Information Modulation for Cross-Modal Representation Learning
    Zeng, Ying
    Mai, Sijie
    Yan, Wenjun
    Hu, Haifeng
    IEEE TRANSACTIONS ON MULTIMEDIA, 2024, 26 : 2178 - 2191
  • [2] Cross-Modal Discrete Representation Learning
    Liu, Alexander H.
    Jin, SouYoung
    Lai, Cheng-I Jeff
    Rouditchenko, Andrew
    Oliva, Aude
    Glass, James
    PROCEEDINGS OF THE 60TH ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL 2022), VOL 1: (LONG PAPERS), 2022, : 3013 - 3035
  • [3] Multimodal Graph Learning for Cross-Modal Retrieval
    Xie, Jingyou
    Zhao, Zishuo
    Lin, Zhenzhou
    Shen, Ying
    PROCEEDINGS OF THE 2023 SIAM INTERNATIONAL CONFERENCE ON DATA MINING, SDM, 2023, : 145 - 153
  • [4] Quaternion Representation Learning for cross-modal matching
    Wang, Zheng
    Xu, Xing
    Wei, Jiwei
    Xie, Ning
    Shao, Jie
    Yang, Yang
    KNOWLEDGE-BASED SYSTEMS, 2023, 270
  • [5] Hybrid representation learning for cross-modal retrieval
    Cao, Wenming
    Lin, Qiubin
    He, Zhihai
    He, Zhiquan
    NEUROCOMPUTING, 2019, 345 : 45 - 57
  • [6] Cross-modal contrastive learning for multimodal sentiment recognition
    Yang, Shanliang
    Cui, Lichao
    Wang, Lei
    Wang, Tao
    APPLIED INTELLIGENCE, 2024, 54 (05) : 4260 - 4276
  • [7] Deep Multimodal Transfer Learning for Cross-Modal Retrieval
    Zhen, Liangli
    Hu, Peng
    Peng, Xi
    Goh, Rick Siow Mong
    Zhou, Joey Tianyi
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2022, 33 (02) : 798 - 810
  • [8] Scalable Deep Multimodal Learning for Cross-Modal Retrieval
    Hu, Peng
    Zhen, Liangli
    Peng, Dezhong
    Liu, Pei
    PROCEEDINGS OF THE 42ND INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL (SIGIR '19), 2019, : 635 - 644
  • [9] Cross-modal contrastive learning for multimodal sentiment recognition
    Shanliang Yang
    Lichao Cui
    Lei Wang
    Tao Wang
    Applied Intelligence, 2024, 54 : 4260 - 4276
  • [10] Disentangled Representation Learning for Cross-Modal Biometric Matching
    Ning, Hailong
    Zheng, Xiangtao
    Lu, Xiaoqiang
    Yuan, Yuan
    IEEE TRANSACTIONS ON MULTIMEDIA, 2022, 24 : 1763 - 1774