Image Representations Learned With Unsupervised Pre-Training Contain Human-like Biases
Image Representations Learned With Unsupervised Pre-Training Contain Human-like Biases
复制标题
通过无监督预训练学习的图像表示包含类人偏差
DOI：
10.1145/3442188.3445932
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Aylin Caliskan
中科院分区：
文献类型：
--
作者：
Ryan Steed;Aylin Caliskan
机器学习的最新进展利用来自网络的大量未标记图像数据集来学习从图像分类到人脸识别的通用图像表示。但是，无监督的计算机视觉模型是否会自动学习隐含模式并嵌入可能产生有害下游影响的社会偏见？我们开发了一种新方法来量化图像中社会概念和属性的表示之间的偏见关联。我们发现，在ImageNet上训练的最先进的无监督模型，ImageNet是一个从互联网图像中策划的流行基准图像数据集，可以自动学习种族，性别和交叉偏见。我们从社会心理学中复制了8个先前记录的人类偏见，从无害的，如昆虫和鲜花，到潜在的有害的，如种族和性别。我们的研究结果与社会心理学关于交叉偏见的三个假设密切相关。在无监督计算机视觉中，我们还首次量化了人类对体重、残疾和几个种族的隐性偏见。当与在线图像数据集中的统计模式进行比较时，我们的研究结果表明，机器学习模型可以自动从人们在网络上被刻板描绘的方式中学习偏见。
Recent advances in machine learning leverage massive datasets of unlabeled images from the web to learn general-purpose image representations for tasks from image classification to face recognition. But do unsupervised computer vision models automatically learn implicit patterns and embed social biases that could have harmful downstream effects? We develop a novel method for quantifying biased associations between representations of social concepts and attributes in images. We find that state-of-the-art unsupervised models trained on ImageNet, a popular benchmark image dataset curated from internet images, automatically learn racial, gender, and intersectional biases. We replicate 8 previously documented human biases from social psychology, from the innocuous, as with insects and flowers, to the potentially harmful, as with race and gender. Our results closely match three hypotheses about intersectional bias from social psychology. For the first time in unsupervised computer vision, we also quantify implicit human biases about weight, disabilities, and several ethnicities. When compared with statistical patterns in online image datasets, our findings suggest that machine learning models can automatically learn bias from the way people are stereotypically portrayed on the web.
登录
查看更多内容
DOI：
10.1073/pnas.1720347115
发表时间：
2018-04-17
影响因子：
11.1
作者：
Garg, Nikhil;Schiebinger, Londa;Zou, James
通讯作者：
Zou, James
影响因子：
7.6
作者：
Greenwald, AG;Nosek, BA;Banaji, MR
通讯作者：
Banaji, MR
影响因子：
7.6
作者：
Greenwald, AG;McGhee, DE;Schwartz, JLK
通讯作者：
Schwartz, JLK
影响因子：
19.5
作者：
Angelina Wang;Alexander Liu;Ryan Zhang;Anat Kleiman;Leslie Kim;Dora Zhao;Iroha Shirai;Arvind Narayanan;Olga Russakovsky
通讯作者：
Angelina Wang;Alexander Liu;Ryan Zhang;Anat Kleiman;Leslie Kim;Dora Zhao;Iroha Shirai;Arvind Narayanan;Olga Russakovsky