Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints

Bias is a problem that will become more common as ML & NLP models in production become more ubiquitous and reach more people. Models inherit the bias of the data they were trained on. We already know that every data, even apparently objective news articles (Bolukbasi et al., 2016) contain significant bias. Zhao et al. propose to inject corpus-level constraints to reduce the bias in structured prediction models in this EMNLP 2017 best paper.


Want to receive more content like this in your inbox?