期刊
IEEE INTERNET OF THINGS JOURNAL
卷 9, 期 9, 页码 6650-6662出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JIOT.2021.3111723
关键词
Deep neural network (DNN) accelerator; DNN model compression; domain-specific accelerator; energy versus performance versus accuracy tradeoff; neural networks
资金
- Italian Ministry of University and Research (MIUR) [ARS01_00345, ARS01_00353]
- University of Catania-Piaceri
This article proposes a compression technique called LineCompress, which aims to reduce memory and communication overhead on resource-constrained IoT devices, thus improving the inference speed and energy efficiency of deep neural networks.
Machine learning techniques, particularly those based on neural networks, are always more often used at the edge of the network by Internet of Things (IoT) nodes. Unfortunately, the computation capabilities demanded by those applications, together with their energy efficiency-related constraints, exceed those exposed by embedded general-purpose processors. For this reason, the use of domain-specific hardware accelerators (DSAs) is considered the most viable solution to the unsustainable Turing tariff of general-purpose hardware. Starting from the observation that memory and communication traffic account for a large fraction of the overall latency and energy in deep neural network (DNN) inferences, this article proposes a new compression technique aimed at: 1) reducing the memory footprint for storing the model parameters of a DNN and 2) improving DNN inference latency and energy on resource-constrained IoT devices. The proposed compression technique, namely, LineCompress, is applied on a set of representative convolutional neural networks (CNNs) for object recognition mapped on a state-of-the-art DSA targeted for resource-constrained IoT devices. We show that on average, 7.4 x memory footprint reduction can be obtained, thus reducing the memory and communication traffic that result to 77% and 87% inference latency and energy reduction, respectively, trading-off efficiency versus accuracy.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据