ISSN 1000-1239 CN 11-1777/TP

计算机研究与发展 ›› 2021, Vol. 58 ›› Issue (12): 2573-2584.doi: 10.7544/issn1000-1239.2021.20210999

所属专题: 2021可解释智能学习方法及其应用专题

• 人工智能 • 上一篇    下一篇

基于对比约束的可解释小样本学习

张玲玲1,陈一苇1,吴文俊1,魏笔凡1,罗炫1,常晓军2,刘均1   

  1. 1(西安交通大学计算机科学与技术学院 西安 710049);2(皇家墨尔本理工大学计算技术学院 澳大利亚墨尔本 3000) (zhanglling@xjtu.edu.cn)
  • 出版日期: 2021-12-01
  • 基金资助: 
    国家重点研发计划项目(2020AAA0108800);国家自然科学基金项目(62137002,61937001,62176209,62176207,62106190,62050194);国家自然科学基金创新群体(61721002);教育部创新团队(IRT_17R86);基于MOOC中国的“一带一路”人才培养的线上线下混合教学支撑信息化平台与服务体系;中国博士后面上项目(2020M683493);中国工程科技知识中心项目;中央高校基本科研项目(xhj032021013-02)

Interpretable Few-Shot Learning with Contrastive Constraint

Zhang Lingling1, Chen Yiwei1, Wu Wenjun1, Wei Bifan1, Luo Xuan1, Chang Xiaojun2, Liu Jun1   

  1. 1(School of Computer Science and Technology, Xian Jiaotong University, Xian 710049)2(School of Computing Technologies, Royal Melbourne Institute of Technology University, Melbourne, Australia 3000)
  • Online: 2021-12-01
  • Supported by: 
    This work was supported by the National Key Research and Development Program of China (2020AAA0108800), the National Natural Science Foundation of China (62137002, 61937001, 62176209, 62176207, 62106190, 62050194), the Innovative Research Group of the National Natural Science Foundation of China (61721002), the Innovation Research Team of Ministry of Education (IRT_17R86), the Consulting Research Project of Chinese Academy of Engineering “The Online and Offline Mixed Educational Service System for ‘The Belt and Road’ Training in MOOC China”, China Postdoctoral Science Foundation (2020M683493), the Project of China Knowledge Centre for Engineering Science and Technology, and the Fundamental Research Funds for the Central Universities (xhj032021013-02).

摘要: 不同于基于大规模监督的深度学习方法,小样本学习旨在从极少的几个样本中学习这类样本的特性,其更符合人脑的视觉认知机制.近年来,小样本学习受到很多学者关注,他们联合元学习训练模式与度量学习理论,挖掘查询集(无标记样本)和支持集(少量标记样本)在特征空间的语义相似距离,取得不错的小样本分类性能.然而,这些方法的可解释性偏弱,不能为用户提供一种便于直观理解的小样本推理过程.为此,提出一种基于区域注意力机制的小样本分类网络INT-FSL,旨在揭示小样本分类中的2个关键问题:1)图像哪些关键位置的视觉特征在决策中发挥了重要作用;2)这些关键位置的视觉特征能体现哪些类别的特性.除此之外,尝试在每个小样本元任务中设计全局和局部2种对比学习机制,利用数据内部信息来缓解小样本场景中的监督信息匮乏问题.在3个真实图像数据集上进行了详细的实验分析,结果表明:所提方法INT-FSL不仅能有效提升当前小样本学习方法的分类性能,还具备良好的过程可解释性.

关键词: 小样本学习, 可解释性分析, 对比学习, 局部描述子, 图像识别

Abstract: Different from deep learning with large scale supervision, few-shot learning aims to learn the samples characteristics from a few labeled examples. Apparently, few-shot learning is more in line with the visual cognitive mechanism of the human brain. In recent years, few-shot learning has attracted more researchers attention. In order to discover the semantic similarities between the query set (unlabeled image) and support set (few labeled images) in feature embedding space, methods which combine meta-learning and metric learning have emerged and achieved great performance on few-shot image classification tasks. However, these methods lack the interpretability, which means they could not provide a reasoning explainable process like human cognitive mechanism. Therefore, we propose a novel interpretable few-shot learning method called INT-FSL based on the positional attention mechanism, which aims to reveal two key problems in few-shot classification: 1)Which parts of the unlabeled image play an important role in classification task; 2)Which class of features reflected by the key parts. Besides, we design the contrastive constraints on global and local levels in every few-shot meta task, for alleviating the limited supervision with the internal information of the data. We conduct extensive experiments on three image benchmark datasets. The results show that the proposed model INT-FSL not only could improve the classification performance on few-shot learning effectively, but also has good interpretability in the reasoning process.

Key words: few-shot learning, interpretable analysis, contrastive learning, local descriptor, image recognition

中图分类号: