Loading Cost-Aware Model Caching and Request Routing for Cooperative Edge Inference | IEEE Conference Publication | IEEE Xplore