salesforce logo About
Research
Ethics
Blog
Outreach
Products
Careers
AI Economist
salesforce logo ☰
About Research Publications Open source Asia Lab Ethics Blog Outreach Products Careers AI Economist About Get Involved Fork us on Github Connect on Slack

14 Jul 2020

It’s Morphin’ Time! Combating Linguistic Discrimination with Inflectional Perturbations

Samson Tan · #research

Morpheus exposes the potential allocative harms of popular pretrained NLP models by simulating inflectional variation. We propose adversarial fine-tuning for mitigating the effects of training only on error-free Standard English data.

30 Jun 2020

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

Tianlu Wang · #research

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.

Page 1 of 1
salesforce logo
Content Terms Privacy Cookie Preferences Press
© Copyright 2020 Salesforce.com, inc. All rights reserved. Rights of ALBERT EINSTEIN are used with permission of The Hebrew University of Jerusalem. Represented exclusively by Greenlight.