Debiasing Embeddings for Reduced Gender Bias in Text Classification

By Flavien Prost, Nithum Thain, Tolga Bolukbasi. 2019

53 citations recorded by the source corpus.

(Bolukbasi et al., 2016) demonstrated that pretrained word embeddings can inherit gender bias from the data they were trained on. We investigate how this bias affects downstream classification tasks, using the case study of occupation classification (De-Arteaga et al., 2019). We show that traditional techniques for debiasing embeddings can actually worsen the bias of the downstream classifier by providing a less noisy channel for communicating gender information. With a relatively minor adjustment, however, we show how these same techniques can be used to simultaneously reduce bias and maintain high classification accuracy.

Subfields: Artificial Intelligence, Natural Language Processing, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

In AI Ethics, Fairness & Society

Citations in the corpus

Language (Technology) is Power: A Critical Survey of “Bias” in NLP (2020, neutral): “…Florez (2019); Jumelet et al. (2019); Lauscher et al. (2019); Liang et al. (2019); Maudslay et al. (2019); May et al. (2019); Prates et al. (2019); Prost et al. (2019); Qian et al. (2019); Swinger et al. (2019); Zhao et al. (2019); Zhou et al. (2019); Ethayarajh (2020); Huang et al. (2020); Jia…”
Measuring and Reducing Gendered Correlations in Pre-trained Models (2020, positive): “Bolukbasi et al. (2016) proposed using linear algebraic techniques to project away dimensions which encode gender, though Prost et al. (2019) found evidence that this method could potentially exacerbate bias for downstream tasks.”
Five sources of bias in natural language processing (2021, background): “For example debiasing embeddings for reducing gender bias in text classification (Prost et al., 2019), dialogue generation (Dinan et al.”
Theories of “Gender” in NLP Bias Research (2022, background): “[110] Flavien Prost, Nithum Thain, and Tolga Bolukbasi.”
Benchmarking Intersectional Biases in NLP (2022, background): “For instance, all four of the aforementioned allocational harm papers measure and/or mitigate gender bias with respect to an NLP-based occupation classifier (De-Arteaga et al., 2019; Prost et al., 2019; Romanov et al., 2019; Zhao et al., 2020).”
Towards Socially Responsible AI: Cognitive Bias-Aware Multi-Objective Learning (2020, background): “Prost, Thain, and Bolukbasi (2019) proposed ways to debias word embedding with an objective to reduce bias in downstream tasks.”
On Transferability of Bias Mitigation Effects in Language Model Fine-Tuning (2021, methodology): “Following Sap et al. (2019), we use an off-the-shelf AAVE dialect predictor (Blodgett et al.”