Equalizing Gender Biases in Neural Machine Translation with Word Embeddings Techniques
This addresses fairness issues in machine translation for users affected by gender stereotypes, though it is incremental as it builds on existing debiasing methods.
The authors tackled gender bias in neural machine translation by integrating debiasing techniques into word embeddings within a Transformer model, achieving up to a one BLEU point improvement on the WMT English-Spanish benchmark and showing effective bias equalization in occupation translations.
Neural machine translation has significantly pushed forward the quality of the field. However, there are remaining big issues with the output translations and one of them is fairness. Neural models are trained on large text corpora which contain biases and stereotypes. As a consequence, models inherit these social biases. Recent methods have shown results in reducing gender bias in other natural language processing tools such as word embeddings. We take advantage of the fact that word embeddings are used in neural machine translation to propose a method to equalize gender biases in neural machine translation using these representations. Specifically, we propose, experiment and analyze the integration of two debiasing techniques over GloVe embeddings in the Transformer translation architecture. We evaluate our proposed system on the WMT English-Spanish benchmark task, showing gains up to one BLEU point. As for the gender bias evaluation, we generate a test set of occupations and we show that our proposed system learns to equalize existing biases from the baseline system.