UM Researchers Unveil 100x More Efficient Edge AI

Researchers at the University of Michigan developed a system that runs AI 100 times more efficiently on edge devices ^[1]. This breakthrough uses a hardware-software co-design to map state space models onto compute-in-memory architectures ^[1]. The design specifically targets local hardware like smartphones and wearable hearing aids ^[1]. This neuromorphic approach drastically lowers power consumption and reduces latency for users ^[1]. By processing data directly on the device, the technology removes the need for constant cloud connectivity ^[1]. This transition allows complex artificial intelligence tasks to function in real-time using small batteries ^[1].