AutoCoMet: Smart Neural Architecture Search via Co-Regulated Shaping Reinforcement
This addresses the problem of efficient model deployment on mobile devices for AI applications, representing an incremental improvement over existing NAS methods.
The paper tackles the challenge of designing deep learning architectures for mobile devices by proposing AutoCoMet, a Neural Architecture Search framework that optimizes for hardware and task contexts, achieving a 3x speedup in search time.
Designing suitable deep model architectures, for AI-driven on-device apps and features, at par with rapidly evolving mobile hardware and increasingly complex target scenarios is a difficult task. Though Neural Architecture Search (NAS/AutoML) has made this easier by shifting paradigm from extensive manual effort to automated architecture learning from data, yet it has major limitations, leading to critical bottlenecks in the context of mobile devices, including model-hardware fidelity, prohibitive search times and deviation from primary target objective(s). Thus, we propose AutoCoMet that can learn the most suitable DNN architecture optimized for varied types of device hardware and task contexts, ~ 3x faster. Our novel co-regulated shaping reinforcement controller together with the high fidelity hardware meta-behavior predictor produces a smart, fast NAS framework that adapts to context via a generalized formalism for any kind of multi-criteria optimization.