Double Hard-Debias: Tailoring Word Embeddings for Gender Bias Mitigation

Word embeddings inherit strong gender bias in data which can be further amplified by downstream models. We propose to purify word embeddings against corpus regularities such as word frequency prior to inferring and removing the gender subspace, which significantly improves the debiasing performance.

30 Jun 2020 • #research

Explaining Solutions to Physical Reasoning Tasks

We show that deep neural models can describe common sense physics in a valid and sufficient way that is also generalizable. Our ESPRIT framework is trained on a new dataset with physics simulations and descriptions that we collected and have open-sourced.

05 May 2020 • #research

ERASER: A Benchmark to Evaluate Rationalized NLP Models

Many NLP applications today deploy state-of-the-art deep neural networks that are essentially black-boxes. One of the goals of Explainable AI (XAI) is to have AI models reveal why and how they make their predictions so that these predictions are interpretable by a human. But work in this direction has been

08 Nov 2019 • #research