Overview
This course focuses on decentralized application-level adaptive scheduling for multi-instance Deep Neural Networks (DNNs) on open mobile devices. The learning outcomes include understanding the challenges of scheduling multiple DNN-powered apps on mobile devices and implementing a decentralized scheduling mechanism using Deep Reinforcement Learning. The course teaches skills such as designing adaptive scheduling solutions, optimizing app performance on mobile systems, and leveraging Deep Reinforcement Learning for scheduling decisions. The teaching method involves presenting research findings, proposing a novel solution, and demonstrating experimental results. The intended audience for this course includes researchers, developers, and practitioners interested in optimizing DNN workloads on open mobile systems.
Syllabus
USENIX ATC '23 - Decentralized Application-Level Adaptive Scheduling for Multi-Instance DNNs on...
Taught by
USENIX