Gender Bias in Contextualized Word Embeddings
arXiv:1904.03310
Abstract
In this paper, we quantify, analyze and mitigate gender bias exhibited in ELMo's contextualized word vectors. First, we conduct several intrinsic analyses and find that (1) training data for ELMo contains significantly more male than female entities, (2) the trained ELMo embeddings systematically encode gender information and (3) ELMo unequally encodes gender information about male and female entities. Then, we show that a state-of-the-art coreference system that depends on ELMo inherits its bias and demonstrates significant bias on the WinoBias probing corpus. Finally, we explore two methods to mitigate such gender bias and show that the bias demonstrated on WinoBias can be eliminated.
References in corpus (5)
- Distributed Representations of Words and Phrases and their Compositionality
- Semantics derived automatically from language corpora contain human-like biases
- Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
- Getting Gender Right in Neural Machine Translation
- Equalizing Gender Biases in Neural Machine Translation with Word Embeddings Techniques
Cited by in corpus (15)
- Think Locally, Act Globally: Federated Learning with Local and Global Representations
- Causal Mediation Analysis for Interpreting Neural NLP: The Case of Gender Bias
- Unmasking Contextual Stereotypes: Measuring and Mitigating BERT's Gender Bias
- Mitigating Gender Bias in Natural Language Processing: Literature Review
- Hurtful Words: Quantifying Biases in Clinical Contextual Word Embeddings
- Two Simple Ways to Learn Individual Fairness Metrics from Data
- Automated Self-Supervised Learning for Graphs
- What's in the Box? A Preliminary Analysis of Undesirable Content in the Common Crawl Corpus
- Does Data Augmentation Improve Generalization in NLP?
- Intersectional Bias in Causal Language Models
- Evaluating Metrics for Bias in Word Embeddings
- Generating Interactive Worlds with Text
- BATS: A Spectral Biclustering Approach to Single Document Topic Modeling and Segmentation
- Adversarial Black-Box Attacks On Text Classifiers Using Multi-Objective Genetic Optimization Guided By Deep Networks
- Mitigating Gender Bias in Captioning Systems