DNN Memory Footprint Reduction via Post-Training Intra-Layer Multi-Precision Quantization

arXiv:2404.02947v1 Announce Type: new
Abstract: The imperative to deploy Deep Neural Network (DNN) models on resource-constrained edge devices, spurred by privacy concerns, has become increasingly apparent. To facilitate the transition from cloud to edge computing, this paper introduces a technique that effectively reduces the memory footprint of DNNs, accommodating the limitations of resource-constrained edge devices while preserving model accuracy. Our proposed technique, named Post-Training Intra-Layer Multi-Precision Quantization (PTILMPQ), employs a post-training quantization approach, eliminating the need for extensive training data. …

abstract arxiv become cloud computing concerns cs.ai cs.lg deep neural network deploy devices dnn edge edge computing edge devices layer limitations memory network neural network paper precision privacy quantization training transition type via