Project Grant 2616836

The National Science Foundation Division of Information and Intelligent Systems awarded the University of South Florida $301,642 on August 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop memory-centric hardware architectures for energy-efficient artificial intelligence inference on edge devices. The project runs through July 31, 2029, and addresses the challenge of deploying AI in power-constrained, compact devices such as wearable health monitors, environmental sensors, and infrastructure monitoring platforms by moving computation closer to memory and using decision-focused data representations that reduce costly numerical operations.

The research comprises three activities: developing encoding and accumulation methods that replace dense multiply-and-accumulate arithmetic with distributed representations, accumulation, and comparison; investigating reliable AI inference using these alternative approaches; and releasing research artifacts for broader research and education use. The work is performed in Tampa, Florida. The project also includes graduate and undergraduate student training in cross-layer hardware and AI design, plus outreach activities to introduce younger students to energy-efficient computing concepts. These advances aim to reduce energy consumption, improve operational reliability, and limit unnecessary data transfers that present security and privacy risks in applications spanning healthcare, environmental monitoring, infrastructure safety, and secure computing.