REDS: Resource-Efficient Deep Subnetworks for Dynamic Resource Constraints

arXiv:2311.13349v2 Announce Type: replace
Abstract: Deep models deployed on edge devices frequently encounter resource variability, which arises from fluctuating energy levels, timing constraints, or prioritization of other critical tasks within the system. State-of-the-art machine learning pipelines generate resource-agnostic models, not capable to adapt at runtime. In this work we introduce Resource-Efficient Deep Subnetworks (REDS) to tackle model adaptation to variable resources. In contrast to the state-of-the-art, REDS use structured sparsity constructively by exploiting permutation invariance of neurons, which allows for …

abstract adapt art arxiv constraints cs.lg devices dynamic edge edge devices energy generate machine machine learning pipelines state tasks type work