NSF HDR ML Anomaly Detection Challenge

The National Science Foundation's (NSF) initiative, Harnessing the Data Revolution (HDR) Big Idea, is a nationwide effort to facilitate innovative modes of data-driven discovery addressing fundamental questions at the frontiers of science and engineering. To raise interest in critical issues within the HDR community, we aim to develop a cross-domain ML challenge for anomaly detection across the five science-focused HDR institutes. This interdisciplinary challenge seeks to provide participants with diverse datasets spanning various scientific domains covering climate science, phylogenetics, materials science, and the identification of astrophysical anomalies from LIGO. The participants should design a novel foundation model for anomaly detection. The algorithm should be able to find anomalies across many different datasets with one metric. We aim to utilize the open-source benchmark ecosystem Codabench to host the challenge and ensure a Findable, Accessible, Interoperable, and Reusable (FAIR) dataset and workflow for the participants from other communities to contribute. The intention is to facilitate contributions from participants across different communities, promoting collaboration and advancing the broader goal of data-driven discovery.