REALTIME

latency-sensitive ML at the edge

This project is funded by the NSF and is about resilience in moble edge cloud (MEC) systems. The idea is that mobile devices can do some machine learning infernes locally but occasionally may need to us a physically close compute center (the MEC) to offload some computation. The work my group is doing is on understanding “when to ask for help” in these systems, using tools from learning theory, information theory, networking, and privacy.

lease see our project page for more details!