To safely deploy deep learning-based computer vision models for computer-aided detection and diagnosis, we must ensure that they are robust and reliable. Towards that goal, algorithmic auditing has received substantial attention. To guide their audit procedures, existing methods rely on heuristic approaches or high-level objectives (e.g., non-discrimination in regards to protected attributes, such as sex, gender, or race). However, algorithms may show bias with respect to various attributes beyond the more obvious ones, and integrity issues related to these more subtle attributes can have serious consequences. To enable the generation of actionable, data-driven hypotheses which identify specific dataset attributes likely to induce model bias, we contribute a first technique for the rigorous, quantitative screening of medical image datasets. Drawing from literature in the causal inference and information theory domains, our procedure decomposes the risks associated with dataset attributes in terms of their detectability and utility (defined as the amount of information knowing the attribute gives about a task label). To demonstrate the effectiveness and sensitivity of our method, we develop a variety of datasets with synthetically inserted artifacts with different degrees of association to the target label that allow evaluation of inherited model biases via comparison of performance against true counterfactual examples. Using these datasets and results from hundreds of trained models, we show our screening method reliably identifies nearly imperceptible bias-inducing artifacts. Lastly, we apply our method to the natural attributes of a popular skin-lesion dataset and demonstrate its success. Our approach provides a means to perform more systematic algorithmic audits and guide future data collection efforts in pursuit of safer and more reliable models.
TabularMark: Watermarking Tabular Datasets for Machine Learning
Zhejiang University(浙江大学);University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);McMaster University(麦克马斯特大学);Tokyo Institute of Technology(东京工业大学);Emory University(埃默里大学)
Detecting Dataset Bias in Medical AI: A Generalized and Modality-Agnostic Auditing Framework
The Johns Hopkins University(约翰霍普金斯大学);The Johns Hopkins University Applied Physics Laboratory(约翰霍普金斯大学应用物理实验室);Center for Devices and Radiological Health, U.S. Food and Drug Administration(美国食品药品监督管理局设备与放射健康中心)
DUCX: Decomposing Unfairness in Tool-Using Chest X-ray Agents
Institute of Artificial Intelligence, Hefei Comprehensive National Science Center, Anhui, China(人工智能研究所,合肥国家科学中心,安徽,中国);The University of British Columbia, Vancouver, BC V6Z 1Z4, Canada(不列颠哥伦比亚大学,温哥华,BC V6Z 1Z4,加拿大);Vector Institute, Toronto, ON M5G 1M1, Canada(向量研究所,多伦多,ON M5G 1M1,加拿大)
Stronger Baseline Models -- A Key Requirement for Aligning Machine Learning Research with Clinical Utility
University of Minnesota(明尼苏达大学);Medical College of Wisconsin(威斯康星医学院);Bud and Sue Selig Hub for Surgical Data Science, Medical College of Wisconsin(威斯康星医学院 Bud 和 Sue Selig 外科数据科学中心);Department of Surgery, Medical College of Wisconsin(威斯康星医学院外科系)
On the Fragility of Data Attribution When Learning Is Distributed
Department of Computer Science and Software Engineering, Auburn University, Auburn, Alabama, USA(计算机科学与软件工程系,阿伯茨温泉大学,阿伯茨温泉,阿拉巴马州,美国);Department of Computer Science, University of Tulsa, Tulsa, Oklahoma, USA(计算机科学系,塔尔萨大学,塔尔萨,俄克拉荷马州,美国);Department of Computer Science and Information Engineering, National Central University, Taoyuan, Taiwan(计算机科学与信息工程系,国立中央大学,桃园,台湾)