Debiasing Sentence Embedders through Contrastive Word Pairs

arXiv:2403.18555v1 Announce Type: new
Abstract: Over the last years, various sentence embedders have been an integral part in the success of current machine learning approaches to Natural Language Processing (NLP). Unfortunately, multiple sources have shown that the bias, inherent in the datasets upon which these embedding methods are trained, is learned by them. A variety of different approaches to remove biases in embeddings exists in the literature. Most of these approaches are applicable to word embeddings and in fewer cases …

abstract arxiv bias cs.cl current datasets embedding integral language language processing machine machine learning multiple natural natural language natural language processing nlp part processing success them through type word