Timezone: »

 
Spotlight
When Does Group Invariant Learning Survive Spurious Correlations?
Yimeng Chen · Ruibin Xiong · Zhi-Ming Ma · Yanyan Lan

Tue Dec 06 05:00 PM -- 07:00 PM (PST) @
in Lightning Talks 2B-3 »

By inferring latent groups in the training data, recent works introduce invariant learning to the case where environment annotations are unavailable. Typically, learning group invariance under a majority/minority split is empirically shown to be effective in improving out-of-distribution generalization on many datasets. However, theoretical guarantee for these methods on learning invariant mechanisms is lacking. In this paper, we reveal the insufficiency of existing group invariant learning methods in preventing classifiers from depending on spurious correlations in the training set. Specifically, we propose two criteria on judging such sufficiency. Theoretically and empirically, we show that existing methods can violate both criteria and thus fail in generalizing to spurious correlation shifts. Motivated by this, we design a new group invariant learning method, which constructs groups with statistical independence tests, and reweights samples by group label proportion to meet the criteria. Experiments on both synthetic and real data demonstrate that the new method significantly outperforms existing group invariant learning methods in generalizing to spurious correlation shifts.

Keywords: spurious correlation · invariant learning · environment inference
[ Slides] 

Author Information

Yimeng Chen (Academy of Mathematics and Systems Science, Chinese Academy of Sciences)
Ruibin Xiong (Institude of Computing Technology, Chinese Academy of Sciences)
Zhi-Ming Ma
Yanyan Lan (Tsinghua University, Tsinghua University)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors