A Converting Autoencoder Toward Low-latency and Energy-efficient DNN Inference at the Edge

arXiv:2403.07036v1 Announce Type: new
Abstract: Reducing inference time and energy usage while maintaining prediction accuracy has become a significant concern for deep neural networks (DNN) inference on resource-constrained edge devices. To address this problem, we propose a novel approach based on "converting" autoencoder and lightweight DNNs. This improves upon recent work such as early-exiting framework and DNN partitioning. Early-exiting frameworks spend different amounts of computation power for different input data depending upon their complexity. However, they can be inefficient in …

abstract accuracy arxiv autoencoder become cs.cv cs.dc cs.lg devices dnn edge edge devices energy inference latency low networks neural networks novel prediction the edge type usage