Counterfactually Measuring and Eliminating Social Bias in Vision-Language Pre-training Models
Authors:
Open Access
- 10 October 2022
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 4996-5004
Abstract
Vision-Language Pre-training (VLP) models have achieved state-of-the-art performance in numerous cross-modal tasks. Since they are optimized to capture the statistical properties of intra- and inter-modality, there remains risk to learn social biases presented in the data as well. In this work, we (1) introduce a counterfactual-based bias measurement CounterBias to quantify the social bias in VLP models by comparing the [MASK]ed prediction probabilities of factual and counterfactual samples; (2) construct a novel VL-Bias dataset including 24K image-text pairs for measuring gender bias in VLP models, from which we observed that significant gender bias is prevalent in VLP models; and (3) propose a VLP debiasing method FairVLP to minimize the difference in the [MASK]ed prediction probabilities between factual and counterfactual image-text pairs for VLP debiasing. Although CounterBias and FairVLP focus on social bias, they are generalizable to serve as tools and provide new insights to probe and regularize more knowledge in VLP models.Keywords
Funding Information
- The National Key R&D Program of China (2018AAA0100604)
This publication has 23 references indexed in Scilit:
- A Corpus for Reasoning about Natural Language Grounded in Photographs2019
- Measuring Bias in Contextualized Word Representations2019
- Gender Bias in Contextualized Word Embeddings2019
- Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods2018
- Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks2017
- Age Progression/Regression by Conditional Adversarial Autoencoder2017
- Deep Learning Face Attributes in the Wild2015
- From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptionsTransactions of the Association for Computational Linguistics, 2014
- Microsoft COCO: Common Objects in Context2014
- Glove: Global Vectors for Word Representation2014