Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints

Abstract

Language is increasingly being used to define rich visual recognition problems with supporting image collections sourced from the web. Structured prediction models are used in these tasks to take advantage of correlations between co-occurring labels and visual input but risk inadvertently encoding social biases found in web corpora. In this work, we study data and models associated with multilabel object classification and visual semantic role labeling. We find that (a) datasets for these tasks contain significant gender bias and (b) models trained on these datasets further amplify existing bias. For example, the activity cooking is over 33% more likely to involve females than males in a training set, and a trained model further amplifies the disparity to 68% at test time. We propose to inject corpus-level constraints for calibrating existing structured prediction models and design an algorithm based on Lagrangian relaxation for collective inference. Our method results in almost no performance loss for the underlying recognition task but decreases the magnitude of bias amplification by 47.5% and 40.5% for multilabel classification and visual semantic role labeling, respectively.

11 pages, published in EMNLP 2017

References in corpus (3)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Microsoft COCO Captions: Data Collection and Evaluation Server
• A Tutorial on Dual Decomposition and Lagrangian Relaxation for Inference in Natural Language Processing

Cited by in corpus (14)

• Towards Explainable Neural-Symbolic Visual Reasoning
• Fairness-Aware Explainable Recommendation over Knowledge Graphs
• Group-Fair Online Allocation in Continuous Time
• Measuring Social Biases of Crowd Workers using Counterfactual Queries
• Exposing and Correcting the Gender Bias in Image Captioning Datasets and Models
• Jigsaw-VAE: Towards Balancing Features in Variational Autoencoders
• Exploring Stereotypes and Biased Data with the Crowd
• Towards Reducing Bias in Gender Classification
• Maximal adversarial perturbations for obfuscation: Hiding certain attributes while preserving rest
• Correcting Exposure Bias for Link Recommendation
• Reducing Overlearning through Disentangled Representations by Suppressing Unknown Tasks
• Towards classification parity across cohorts
• TFW, DamnGina, Juvie, and Hotsie-Totsie: On the Linguistic and Social Aspects of Internet Slang
• Adversarial Examples Generation for Reducing Implicit Gender Bias in Pre-trained Models