Project Grant 2310482

This Project Grant award from the National Science Foundation's Division of Information and Intelligent Systems (CFDA 47.070 Computer and Information Science and Engineering) provides $300,000 over 3 years to Carnegie Mellon University (CMU) to develop fair outlier detection algorithms. The project aims to encode fairness into data mining and deep learning-based outlier detection methods, which are commonly used in applications like fraud detection, surveillance, and policing. The key objectives are to define fairness metrics for outlier detection, explore techniques to audit algorithms for unfairness and post-process outputs to improve fairness, and directly incorporate fairness into deep learning-based outlier detection models. The research will be evaluated on social media platform content filtering and medical imaging data preprocessing use cases. CMU, a leading research institution, will collaborate with industry and academic partners to deliver these capabilities, which have the potential to make outlier detection more equitable across protected status groups.