This Project Grant from the National Science Foundation's Division of Information and Intelligent Systems, under the Computer and Information Science and Engineering program (CFDA 47.070), provides $246,643 to Penn State University for research on detecting and mitigating non-functional vulnerabilities in machine learning libraries. Specifically, the university will develop methods to detect denial-of-service vulnerabilities and quantify side-channel vulnerabilities in core ML libraries such as NumPy and TensorFlow. It will also explore causally guided statistical techniques to debug these issues by localizing root causes and guiding mitigation policies. As part of broader impact objectives, the university aims to increase participation from underrepresented groups in security research through special topic courses, undergraduate research experiences, and K-12 summer camps. The three-year award period runs from April 1, 2023 through March 31, 2026.
Generated 1/6/24, 12:06 PM