Deep Stable Learning for Out-Of-Distribution Generalization

被引:183
作者
Zhang, Xingxuan [1 ]
Cui, Peng [1 ,2 ]
Xu, Renzhe [1 ]
Zhou, Linjun [1 ]
He, Yue [1 ]
Shen, Zheyan [1 ]
机构
[1] Tsinghua Univ, Dept Comp Sci, Beijing, Peoples R China
[2] Beijing Key Lab Networked Multimedia, Beijing, Peoples R China
来源
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 | 2021年
基金
中国国家自然科学基金; 国家重点研发计划;
关键词
D O I
10.1109/CVPR46437.2021.00533
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Approaches based on deep neural networks have achieved striking performance when testing data and training data share similar distribution, but can significantly fail otherwise. Therefore, eliminating the impact of distribution shifts between training and testing data is crucial for building performance-promising deep models. Conventional methods assume either the known heterogeneity of training data (e.g. domain labels) or the approximately equal capacities of different domains. In this paper, we consider a more challenging case where neither of the above assumptions holds. We propose to address this problem by removing the dependencies between features via learning weights for training samples, which helps deep models get rid of spurious correlations and, in turn, concentrate more on the true connection between discriminative features and labels. Extensive experiments clearly demonstrate the effectiveness of our method on multiple distribution generalization benchmarks compared with state-of-the-art counterparts. Through extensive experiments on distribution generalization benchmarks including PACS, VLCS, MNIST-M, and NICO, we show the effectiveness of our method compared with state-of-the-art counterparts.
引用
收藏
页码:5368 / 5378
页数:11
相关论文
共 64 条
[1]  
Adebayo J, 2018, ADV NEUR IN, V31
[2]  
[Anonymous], 2020, AAAI
[3]  
[Anonymous], 2019, arXiv
[4]  
[Anonymous], 2020, UNCERTAINTY ARTIFICI
[5]  
Arjovsky Martin, 2019, CoRR, P1
[6]  
Bahng, 2019, LEARNING DEBIASED RE
[7]  
Balaji Y, 2018, ADV NEUR IN, V31
[8]   Domain Generalization by Solving Jigsaw Puzzles [J].
Carlucci, Fabio M. ;
D'Innocente, Antonio ;
Bucci, Silvia ;
Caputo, Barbara ;
Tommasi, Tatiana .
2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, :2224-2233
[9]  
Deng J, 2009, PROC CVPR IEEE, P248, DOI 10.1109/CVPRW.2009.5206848
[10]   Deep Domain Generalization With Structured Low-Rank Constraint [J].
Ding, Zhengming ;
Fu, Yun .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2018, 27 (01) :304-313