Supervised Compression for Resource-Constrained Edge Computing Systems

被引:19
|
作者
Matsubara, Yoshitomo [1 ]
Yang, Ruihan [1 ]
Levorato, Marco [1 ]
Mandt, Stephan [1 ]
机构
[1] Univ Calif Irvine, Dept Comp Sci, Irvine, CA 92717 USA
来源
2022 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV 2022) | 2022年
基金
美国国家科学基金会;
关键词
D O I
10.1109/WACV51458.2022.00100
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
There has been much interest in deploying deep learning algorithms on low-powered devices, including smartphones, drones, and medical sensors. However, full-scale deep neural networks are often too resource-intensive in terms of energy and storage. As a result, the bulk part of the machine learning operation is therefore often carried out on an edge server, where the data is compressed and transmitted. However, compressing data (such as images) leads to transmitting information irrelevant to the supervised task. Another popular approach is to split the deep network between the device and the server while compressing intermediate features. To date, however, such split computing strategies have barely outperformed the aforementioned naive data compression baselines due to their inefficient approaches to feature compression. This paper adopts ideas from knowledge distillation and neural image compression to compress intermediate feature representations more efficiently. Our supervised compression approach uses a teacher model and a student model with a stochastic bottleneck and learnable prior for entropy coding (Entropic Student). We compare our approach to various neural image and feature compression baselines in three vision tasks and found that it achieves better supervised rate-distortion performance while maintaining smaller end-to-end latency. We furthermore show that the learned feature representations can be tuned to serve multiple downstream tasks.
引用
收藏
页码:923 / 933
页数:11
相关论文
共 50 条
  • [1] FedComp: A Federated Learning Compression Framework for Resource-Constrained Edge Computing Devices
    Wu, Donglei
    Yang, Weihao
    Jin, Haoyu
    Zou, Xiangyu
    Xia, Wen
    Fang, Binxing
    IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, 2024, 43 (01) : 230 - 243
  • [2] Adaptive Asynchronous Federated Learning in Resource-Constrained Edge Computing
    Liu, Jianchun
    Xu, Hongli
    Wang, Lun
    Xu, Yang
    Qian, Chen
    Huang, Jinyang
    Huang, He
    IEEE TRANSACTIONS ON MOBILE COMPUTING, 2023, 22 (02) : 674 - 690
  • [3] Knowledge Distillation in Object Detection for Resource-Constrained Edge Computing
    Setyanto, Arief
    Sasongko, Theopilus Bayu
    Fikri, Muhammad Ainul
    Ariatmanto, Dhani
    Agastya, I. Made Artha
    Rachmanto, Rakandhiya Daanii
    Ardana, Affan
    Kim, In Kee
    IEEE ACCESS, 2025, 13 : 18200 - 18214
  • [4] Resource-Constrained Serial Task Offload Strategy in Mobile Edge Computing
    Liu W.
    Huang Y.-C.
    Du W.
    Wang W.
    Ruan Jian Xue Bao/Journal of Software, 2020, 31 (06): : 1889 - 1908
  • [5] Computation Offloading in Resource-Constrained Multi-Access Edge Computing
    Li, Kexin
    Wang, Xingwei
    He, Qiang
    Wang, Jielei
    Li, Jie
    Zhan, Siyu
    Lu, Guoming
    Dustdar, Schahram
    IEEE TRANSACTIONS ON MOBILE COMPUTING, 2024, 23 (11) : 10665 - 10677
  • [6] Adaptive Batch Size for Federated Learning in Resource-Constrained Edge Computing
    Ma, Zhenguo
    Xu, Yang
    Xu, Hongli
    Meng, Zeyu
    Huang, Liusheng
    Xue, Yinxing
    IEEE TRANSACTIONS ON MOBILE COMPUTING, 2023, 22 (01) : 37 - 53
  • [7] To Compute or Not to Compute? Adaptive Smart Sensing in Resource-Constrained Edge Computing
    Ballotta, Luca
    Peserico, Giovanni
    Zanini, Francesco
    Dini, Paolo
    IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING, 2024, 11 (01): : 736 - 749
  • [8] Computation Off-Loading in Resource-Constrained Edge Computing Systems Based on Deep Reinforcement Learning
    Luo, Chuanwen
    Zhang, Jian
    Cheng, Xiaolu
    Hong, Yi
    Chen, Zhibo
    Xing, Xiaoshuang
    IEEE TRANSACTIONS ON COMPUTERS, 2024, 73 (01) : 109 - 122
  • [9] Head Network Distillation: Splitting Distilled Deep Neural Networks for Resource-Constrained Edge Computing Systems
    Matsubara, Yoshitomo
    Callegaro, Davide
    Baidya, Sabur
    Levorato, Marco
    Singh, Sameer
    IEEE ACCESS, 2020, 8 (08) : 212177 - 212193
  • [10] Data cube-based storage optimization for resource-constrained edge computing
    Gao, Liyuan
    Li, Wenjing
    Ma, Hongyue
    Liu, Yumin
    Li, Chunyang
    HIGH-CONFIDENCE COMPUTING, 2024, 4 (04):