Probabilistic Bias Mitigation in Word Embeddings

Hailey Joren, David Alvarez-Melis

It has been shown that word embeddings derived from large corpora tend to incorporate biases present in their training data. Various methods for mitigating these biases have been p…