Project Grant 2312205

The National Science Foundation (NSF) awarded a $237,028 Project Grant to New York University (NYU) under the Computer and Information Science and Engineering program (CFDA 47.070) to develop statistical and algorithmic foundations for robust policy learning in uncertain environments. The goal is to create provably efficient techniques for learning optimization policies that can be deployed in practical settings where the training and operational environments differ, such as when using digital twins or simulators. The research program seeks to establish information-theoretic limits on distributionally robust policy learning, develop efficient estimation schemes for assessing policy performance under distributional shifts, and translate these advances into practical policy learning algorithms. This NSF award reflects the agency's mission to support fundamental research that can transform engineering, scientific, and societal applications through data-driven optimization and decision-making.