`

Timezone: »

 
Poster
Debiased Contrastive Learning
Ching-Yao Chuang · Joshua Robinson · Yen-Chen Lin · Antonio Torralba · Stefanie Jegelka

Mon Dec 07 09:00 PM -- 11:00 PM (PST) @ Poster Session 0 #18

A prominent technique for self-supervised representation learning has been to contrast semantically similar and dissimilar pairs of samples. Without access to labels, dissimilar (negative) points are typically taken to be randomly sampled datapoints, implicitly accepting that these points may, in reality, actually have the same label. Perhaps unsurprisingly, we observe that sampling negative examples from truly different labels improves performance, in a synthetic setting where labels are available. Motivated by this observation, we develop a debiased contrastive objective that corrects for the sampling of same-label datapoints, even without knowledge of the true labels. Empirically, the proposed objective consistently outperforms the state-of-the-art for representation learning in vision, language, and reinforcement learning benchmarks. Theoretically, we establish generalization bounds for the downstream classification task.

Author Information

Ching-Yao Chuang (MIT)
Joshua Robinson (MIT)
Yen-Chen Lin (MIT)
Antonio Torralba (MIT)
Stefanie Jegelka (MIT)

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

More from the Same Authors