This $499,237 National Science Foundation project grant supports research at the University of Michigan from October 1, 2022 to September 30, 2025 under the Computer and Information Science and Engineering program (CFDA 47.070).
The research aims to develop a theoretical framework for algorithmic and statistical fraud detection applicable across domains. The investigator will design methods with formal guarantees on efficiency and efficacy, studying abstract fraud detection games and applying statistical analysis to existing algorithmic challenges. Initial goals include analyzing coin flipping, random walk, allocation, and leader election games from the lens of fraud detection. The investigator will also design an optimal-resilience Byzantine agreement protocol using statistical tests to identify misbehaving processes in distributed systems and apply fraud detection concepts to protect randomized data structures with partially exposed internal randomness.
Generated 1/7/24, 4:53 AM