Edge-Host Collaborative Deep Learning

Based on the trade-off study between energy, accuracy, and throughput of an edge platform with an embedded DNN, we are currently exploring collaborative DNN inference between the edge and the host to optimally utilize the available bandwidth and energy.