Project Grant 2616633

The National Science Foundation Division of Computing and Communication Foundations awarded the University of Wisconsin - Madison $249,999 on August 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop graph-centric multi-agent systems powered by large language models.

The project develops new graph-based methods for designing, executing, and adapting multi-agent artificial intelligence systems in which multiple agents communicate and collaborate to solve complex problems. Research focuses on three complementary directions: reliability-aware methods that detect unreliable or adversarial agents, estimate uncertainty in agent communication, and contain the spread of incorrect information through the collaboration graph; efficiency-oriented methods that reduce unnecessary communication and computation by identifying redundant agents, edges, and messages; and adaptive methods that enable systems to respond to unexpected changes in agent availability, interaction structure, or task objectives. The work addresses limitations in current multi-agent systems that rely on static, manually designed communication patterns, making them vulnerable to deployment challenges in real-world environments. The project supports student training in machine learning, large language models, and agentic artificial intelligence.

Work is performed in Madison, Wisconsin, over a period of performance from August 1, 2026, through July 31, 2029. The award is structured as a project grant under the Computer and Information Science and Engineering program.