OntoZSL: Ontology-enhanced Zero-shot Learning

被引:34
|
作者
Geng, Yuxia [1 ]
Chen, Jiaoyan [2 ]
Chen, Zhuo [1 ]
Pan, Jeff Z. [3 ]
Ye, Zhiquan [4 ]
Yuan, Zonggang [5 ]
Jia, Yantao [6 ]
Chen, Huajun [7 ,8 ]
机构
[1] Zhejiang Univ, Hangzhou, Peoples R China
[2] Univ Oxford, Oxford, England
[3] Univ Edinburgh, Edinburgh, Midlothian, Scotland
[4] Zhejiang Univ, Coll Comp Sci, Hangzhou, Peoples R China
[5] Huawei Technol Co Ltd, NAIE CTO Off, Nanjing, Peoples R China
[6] Huawei Technol Co Ltd, Poisson Lab, Beijing, Peoples R China
[7] Zhejiang Univ, Coll Comp Sci, AZFT Knowledge Engine Lab, Hangzhou, Peoples R China
[8] Zhejiang Univ, HIC, AZFT Knowledge Engine Lab, Hangzhou, Peoples R China
关键词
Zero-shot Learning; Ontology; Generative Adversarial Networks; Image Classification; Knowledge Graph Completion; CLASSIFICATION;
D O I
10.1145/3442381.3450042
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Zero-shot Learning (ZSL), which aims to predict for those classes that have never appeared in the training data, has arisen hot research interests. The key of implementing ZSL is to leverage the prior knowledge of classes which builds the semantic relationship between classes and enables the transfer of the learned models (e.g., features) from training classes (i.e., seen classes) to unseen classes. However, the priors adopted by the existing methods are relatively limited with incomplete semantics. In this paper, we explore richer and more competitive prior knowledge to model the inter-class relationship for ZSL via ontology-based knowledge representation and semantic embedding. Meanwhile, to address the data imbalance between seen classes and unseen classes, we developed a generative ZSL framework with Generative Adversarial Networks (GANs). Our main findings include: (i) an ontology-enhanced ZSL framework that can be applied to different domains, such as image classification (IMGC) and knowledge graph completion (KGC); (ii) a comprehensive evaluation with multiple zero-shot datasets from different domains, where our method often achieves better performance than the state-of-the-art models. In particular, on four representative ZSL baselines of IMGC, the ontology-based class semantics outperform the previous priors e.g., the word embeddings of classes by an average of 12.4 accuracy points in the standard ZSL across two example datasets (see Figure 4).
引用
收藏
页码:3325 / 3336
页数:12
相关论文
共 50 条
  • [41] Zero-Shot Program Representation Learning
    Cui, Nan
    Jiang, Yuze
    Gu, Xiaodong
    Shen, Beijun
    arXiv, 2022,
  • [42] Zero-shot Learning With Fuzzy Attribute
    Liu, Chongwen
    Shang, Zhaowei
    Tang, Yuan Yan
    2017 3RD IEEE INTERNATIONAL CONFERENCE ON CYBERNETICS (CYBCONF), 2017, : 277 - 282
  • [43] Detecting Errors with Zero-Shot Learning
    Wu, Xiaoyu
    Wang, Ning
    ENTROPY, 2022, 24 (07)
  • [44] Landmark Selection for Zero-shot Learning
    Guo, Yuchen
    Ding, Guiguang
    Han, Jungong
    Yan, Chenggang
    Zhang, Jiyong
    Dai, Qionghai
    PROCEEDINGS OF THE TWENTY-EIGHTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2019, : 2435 - 2441
  • [45] Recent Advances in Zero-Shot Learning
    Lan Hong
    Fang Zhiyu
    JOURNAL OF ELECTRONICS & INFORMATION TECHNOLOGY, 2020, 42 (05) : 1188 - 1200
  • [46] Learning Attention Propagation for Compositional Zero-Shot Learning
    Khan, Muhammad Gul Zain Ali
    Naeem, Muhammad Ferjad
    Van Gool, Luc
    Pagani, A.
    Stricker, Didier
    Afzal, Muhammad Zeshan
    2023 IEEE/CVF WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV), 2023, : 3817 - 3826
  • [47] Learning MLatent Representations for Generalized Zero-Shot Learning
    Ye, Yalan
    Pan, Tongjie
    Luo, Tonghoujun
    Li, Jingjing
    Shen, Heng Tao
    IEEE TRANSACTIONS ON MULTIMEDIA, 2023, 25 : 2252 - 2265
  • [48] Meta-Learning for Generalized Zero-Shot Learning
    Verma, Vinay Kumar
    Brahma, Dhanajit
    Rai, Piyush
    THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 6062 - 6069
  • [49] Learning the Compositional Domains for Generalized Zero-shot Learning
    Dong, Hanze
    Fu, Yanwei
    Hwang, Sung Ju
    Sigal, Leonid
    Xue, Xiangyang
    COMPUTER VISION AND IMAGE UNDERSTANDING, 2022, 221
  • [50] Zero-shot Learning via Simultaneous Generating and Learning
    Yu, Hyeonwoo
    Lee, Beomhee
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 32 (NIPS 2019), 2019, 32