Joint Multiclass Debiasing of Word Embeddings
This addresses bias in textual data representations for AI applications, offering a more comprehensive solution than single-dimension methods, though it builds incrementally on existing debiasing efforts.
The paper tackles the problem of bias in word embeddings by developing a joint multiclass debiasing approach that can handle multiple bias dimensions simultaneously, such as religion, gender, and race, and shows it can reduce or eliminate bias while preserving meaningful vector relationships.
Bias in Word Embeddings has been a subject of recent interest, along with efforts for its reduction. Current approaches show promising progress towards debiasing single bias dimensions such as gender or race. In this paper, we present a joint multiclass debiasing approach that is capable of debiasing multiple bias dimensions simultaneously. In that direction, we present two approaches, HardWEAT and SoftWEAT, that aim to reduce biases by minimizing the scores of the Word Embeddings Association Test (WEAT). We demonstrate the viability of our methods by debiasing Word Embeddings on three classes of biases (religion, gender and race) in three different publicly available word embeddings and show that our concepts can both reduce or even completely eliminate bias, while maintaining meaningful relationships between vectors in word embeddings. Our work strengthens the foundation for more unbiased neural representations of textual data.