Josh Millar has just released the latest survey paper he lead on energy-aware approaches to optimise deep-learning training and inference on embedded devices, such as those benchmarked in "Benchmarking Ultra-Low-Power µNPUs" recently. This comprehensive survey is particularly relevant for IoT and mobile devices that face substantial energy constraints to prolong battery life or operate intermittently via energy harvesting. Josh synthesizes the evolving landscape of energy-aware deep learning approaches, examining their methodologies, implications for energy consumption and system-level efficiency, and their limitations across different network types, hardware platforms, and application scenarios. The work is especially timely given the push to run more AI workloads on edge devices for privacy, latency, and cost reasons.
We present an overview of such approaches, outlining their methodologies, implications for energy consumption and system-level efficiency, and their limitations in terms of supported network types, hardware platforms, and application scenarios. We hope our review offers a clear synthesis of the evolving energy-aware DL landscape and serves as a foundation for future research in energy-constrained computing.
Any comments, please do let any of us know!