Gender Slopes: Counterfactual Fairness for Computer Vision Models by Attribute Manipulation
Gender Slopes: Counterfactual Fairness for Computer Vision Models by Attribute Manipulation
复制标题
DOI：
10.1145/3422841.3423533
复制
发表时间：
2020-05
期刊：
影响因子：
--
通讯作者：
Jungseock Joo;Kimmo Kärkkäinen
中科院分区：
文献类型：
--
作者：
Jungseock Joo;Kimmo Kärkkäinen
自动计算机视觉系统已经应用于许多领域，包括安全、执法和个人设备，但最近的报告表明，这些系统可能会产生有偏见的结果，歧视某些人口群体的人。然而，诊断和理解模型偏差的潜在真正原因是一项具有挑战性的任务，因为现代计算机视觉系统依赖于复杂的黑箱模型，其行为难以解码。我们建议使用为图像属性操作开发的编码器-解码器网络来合成在性别和种族维度上不同的面部图像，同时保持其他信号不变。我们使用这些合成图像来衡量商业计算机视觉分类器的反事实公平性，通过检查这些分类器受图像中控制的性别和种族线索的影响程度，例如，女性面孔可能会在护士概念上获得更高的分数，而在stem相关概念上获得更低的分数。
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.