This three-year, $600,000 project grant from the National Science Foundation's Computer and Information Science and Engineering program aims to develop learning-augmented mechanisms for algorithmic mechanism design. The grant supports research at Columbia University to design robust mechanisms guided by machine-learned predictions of agents' preferences. Researchers will consider learning-augmented mechanism design for canonical problems in areas like auction design, mechanism design without monetary transfers, online mechanism design, and decentralized settings evaluated over Nash equilibria. The goal is to overcome overly pessimistic impossibility results and enable improved performance guarantees with respect to social welfare and revenue objectives. By incorporating predictions of strategic agents' incentives and valuations, this work has the potential to transform the field of mechanism design.
Generated 1/7/24, 5:10 AM