Tianlu Wang

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

June 30, 2020

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.

AI Research

Enter a valid e-mail address

Select your Country

Select a state/province

Yes, I would like to receive the Salesforce 360 Highlights newsletter as well as marketing emails regarding Salesforce products, services, and events. I can unsubscribe at any time.

I agree to the Privacy Statement and to the handling of my personal information. In particular, I consent to the transfer of my personal information to other countries, including the United States, for the purpose of hosting and processing the information as set forth in the Privacy Statement. Learn More

I understand that these countries may not have the same data protection laws as the country from which I provide my personal information. For more information, click here.

Please read and agree to the Master Subscription Agreement

By registering, you confirm that you agree to the processing of your personal data by Salesforce as described in the Privacy Statement.