Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigation
Authors:
- 1 June 2020
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 8916-8925
Abstract
Computer vision models learn to perform a task by capturing relevant statistics from training data. It has been shown that models learn spurious age, gender, and race correlations when trained for seemingly unrelated tasks like activity recognition or image captioning. Various mitigation techniques have been presented to prevent models from utilizing or learning such biases. However, there has been little systematic comparison between these techniques. We design a simple but surprisingly effective visual recognition benchmark for studying bias mitigation. Using this benchmark, we provide a thorough analysis of a wide range of techniques. We highlight the shortcomings of popular adversarial training approaches for bias mitigation, propose a simple but similarly effective alternative to the inference-time Reducing Bias Amplification method of Zhao et al., and design a domain-independent training technique that outperforms all other methods. Finally, we validate our findings on the attribute classification task in the CelebA dataset, where attribute presence is known to be correlated with the gender of people in the image, and demonstrate that the proposed technique is effective at mitigating real-world gender bias.Keywords
All Related Versions
This publication has 21 references indexed in Scilit:
- Semantics derived automatically from language corpora contain human-like biasesScience, 2017
- Deep Residual Learning for Image Recognition2016
- Deep Learning Face Attributes in the Wild2015
- Simultaneous Deep Transfer Across Domains and Tasks2015
- Classifier adaptation at prediction time2015
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Fairness through awareness2012
- Diagnosing Error in Object Detectors2012
- Discrimination-aware data mining2008
- Adjusting the Outputs of a Classifier to New aPrioriProbabilities: A Simple ProcedureNeural Computation, 2002