Project Grant 2154711

Cornell University was awarded a three-year, $384,616 Project Grant from the National Science Foundation's Computer and Information Science and Engineering program to develop new reinforcement learning algorithms and personalized voice navigation systems. The goal of the research is to create reinforcement learning techniques that can learn efficiently using minimal training data, which could enable applications where data is expensive to collect, such as autonomous vehicles adapting quickly to new road conditions with fewer mistakes. Key activities under the award include developing computationally efficient algorithms for large-scale Markov decision processes involving high-dimensional, complex data; incorporating representation learning into reinforcement learning to allow extraction of compact information from unstructured data; and designing personalized voice navigation systems that can safely and rapidly adapt to individual end-users through sample efficient offline and online reinforcement learning. The award period runs from October 1, 2022 to September 30, 2025.