Adversarial Filters of Dataset Biases

Ronan Le Bras, Swabha Swayamdipta, Chandra Bhagavatula, Rowan Zellers, Matthew Peters, Ashish Sabharwal, Yejin Choi,

Abstract Paper

Wed Jul 15 8 a.m. PDT [iCal] [ Join Zoom ]
Wed Jul 15 8 p.m. PDT [iCal] [ Join Zoom ]
Please do not share or post zoom links


Large neural models have demonstrated human-level performance on language and vision benchmarks, while their performance degrades considerably on adversarial or out-of-distribution samples. This raises the question of whether these models have learned to solve a dataset rather than the underlying task by overfitting to spurious dataset biases. We investigate one recently proposed approach, AFLITE, which adversarially filters such dataset biases, as a means to mitigate the prevalent overestimation of machine performance. We provide a theoretical understanding for AFLITE, by situating it in the generalized framework for optimum bias reduction. We present extensive supporting evidence that AFLITE is broadly applicable for reduction of measurable dataset biases, and that models trained on the filtered datasets yield better generalization to out-of-distribution tasks. Finally, filtering results in a large drop in model performance (e.g., from 92% to 62% for SNLI), while human performance still remains high. Our work thus shows that such filtered datasets can pose new research challenges for robust generalization by serving as upgraded benchmarks.

Chat is not available.