Gender Slopes: Counterfactual Fairness for Computer Vision Models by Attribute Manipulation

Automated computer vision systems have been applied in many domains including security, law enforcement, and personal devices, but recent reports suggest that these systems may produce biased results, discriminating against people in certain demographic groups. Diagnosing and understanding the underlying true causes of model biases, however, are challenging tasks because modern computer vision systems rely on complex black-box models whose behaviors are hard to decode. We propose to use an encoder-decoder network developed for image attribute manipulation to synthesize facial images varying in the dimensions of gender and race while keeping other signals intact. We use these synthesized images to measure counterfactual fairness of commercial computer vision classifiers by examining the degree to which these classifiers are affected by gender and racial cues controlled in the images, e.g., feminine faces may elicit higher scores for the concept of nurse and lower scores for STEM-related concepts.

[1]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[2]  Ryan Cotterell,et al.  Counterfactual Data Augmentation for Mitigating Gender Stereotypes in Languages with Rich Morphology , 2019, ACL.

[3]  Shiguang Shan,et al.  AttGAN: Facial Attribute Editing by Only Changing What You Want , 2017, IEEE Transactions on Image Processing.

[4]  R. Hofmann-Wellenhof,et al.  Comparison of the accuracy of human readers versus machine-learning algorithms for pigmented skin lesion classification: an open, web-based, international, diagnostic study. , 2019, The Lancet. Oncology.

[5]  Bernhard Schölkopf,et al.  Avoiding Discrimination through Causal Reasoning , 2017, NIPS.

[6]  Honglak Lee,et al.  Attribute2Image: Conditional Image Generation from Visual Attributes , 2015, ECCV.

[7]  Krishna P. Gummadi,et al.  Fairness Constraints: Mechanisms for Fair Classification , 2015, AISTATS.

[8]  Kimmo Kärkkäinen,et al.  FairFace: Face Attribute Dataset for Balanced Race, Gender, and Age , 2019, ArXiv.

[9]  Emily Denton,et al.  Detecting Bias with Generative Counterfactual Face Attribute Augmentation , 2019, ArXiv.

[10]  Sharad Goel,et al.  The Measure and Mismeasure of Fairness: A Critical Review of Fair Machine Learning , 2018, ArXiv.

[11]  Ming Yang,et al.  DeepFace: Closing the Gap to Human-Level Performance in Face Verification , 2014, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[12]  Catherine Tucker,et al.  Algorithmic bias? An empirical study into apparent gender-based discrimination in the display of STEM career ads , 2019 .

[13]  Brian Kingsbury,et al.  Understanding Unequal Gender Classification Accuracy from Face Images , 2018, ArXiv.

[14]  Trevor Darrell,et al.  Women also Snowboard: Overcoming Bias in Captioning Models , 2018, ECCV.

[15]  Toniann Pitassi,et al.  Learning Fair Representations , 2013, ICML.

[16]  Matt J. Kusner,et al.  Counterfactual Fairness , 2017, NIPS.

[17]  Vidya Muthukumar,et al.  Color-Theoretic Experiments to Understand Unequal Gender Classification Accuracy From Face Images , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[18]  Catherine E. Tucker,et al.  Algorithmic Bias? An Empirical Study of Apparent Gender-Based Discrimination in the Display of STEM Career Ads , 2019, Manag. Sci..

[19]  Pietro Perona,et al.  Microsoft COCO: Common Objects in Context , 2014, ECCV.

[20]  Timnit Gebru,et al.  Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification , 2018, FAT.

[21]  Guillaume Lample,et al.  Fader Networks: Manipulating Images by Sliding Attributes , 2017, NIPS.

[22]  Jungseock Joo,et al.  Understanding the Political Ideology of Legislators from Social Media Images , 2019, ICWSM.

[23]  T. Brennan,et al.  Evaluating the Predictive Validity of the Compas Risk and Needs Assessment System , 2009 .

[24]  Gang Yu,et al.  BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation , 2018, ECCV.

[25]  Jieyu Zhao,et al.  Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints , 2017, EMNLP.

[26]  Toniann Pitassi,et al.  Fairness through awareness , 2011, ITCS '12.

[27]  Gang Hua,et al.  CVAE-GAN: Fine-Grained Image Generation through Asymmetric Training , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[28]  Xiaogang Wang,et al.  Deep Learning Face Attributes in the Wild , 2014, 2015 IEEE International Conference on Computer Vision (ICCV).

[29]  Ming-Hsuan Yang,et al.  Learning Gender with Support Faces , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[30]  Daniel McDuff,et al.  Characterizing Bias in Classifiers using Generative Models , 2019, NeurIPS.

[31]  Nathan Srebro,et al.  Equality of Opportunity in Supervised Learning , 2016, NIPS.

[32]  Jahna Otterbacher,et al.  Fairness in Proprietary Image Tagging Algorithms: A Cross-Platform Audit on People Images , 2019, ICWSM.

[33]  G. Corrado,et al.  End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography , 2019, Nature Medicine.

[34]  Larry S. Davis,et al.  Explicit Bias Discovery in Visual Question Answering Models , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[35]  Eric Goldman,et al.  Search Engine Bias and the Demise of Search Engine Utopianism , 2006 .