FedPAQ: A Communication-Efficient Federated Learning Method with Periodic Averaging and Quantization

被引:0
|
作者
Reisizadeh, Amirhossein [1 ]
Mokhtari, Aryan [2 ]
Hassani, Hamed [3 ]
Jadbabaie, Ali [4 ]
Pedarsani, Ramtin [1 ]
机构
[1] UC Santa Barbara, Santa Barbara, CA 93106 USA
[2] UT Austin, Austin, TX USA
[3] UPenn, Pittsburgh, PA USA
[4] MIT, Cambridge, MA 02139 USA
关键词
GRADIENT DESCENT;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Federated learning is a distributed framework according to which a model is trained over a set of devices, while keeping data localized. This framework faces several systems-oriented challenges which include (i) communication bottleneck since a large number of devices upload their local updates to a parameter server, and (ii) scalability as the federated network consists of millions of devices. Due to these systems challenges as well as issues related to statistical heterogeneity of data and privacy concerns, designing a provably efficient federated learning method is of significant importance yet it remains challenging. In this paper, we present FedPAQ, a communication-efficient Federated Learning method with Periodic Averaging and Quantization. FedPAQ relies on three key features: (1) periodic averaging where models are updated locally at devices and only periodically averaged at the server; (2) partial device participation where only a fraction of devices participate in each round of the training; and (3) quantized message-passing where the edge nodes quantize their updates before uploading to the parameter server. These features address the communications and scalability challenges in federated learning. We also show that FedPAQ achieves near-optimal theoretical guarantees for strongly convex and non-convex loss functions and empirically demonstrate the communication-computation tradeoff provided by our method.
引用
收藏
页码:2021 / 2030
页数:10
相关论文
共 50 条
  • [1] Communication-Efficient Federated Learning with Adaptive Quantization
    Mao, Yuzhu
    Zhao, Zihao
    Yan, Guangfeng
    Liu, Yang
    Lan, Tian
    Song, Linqi
    Ding, Wenbo
    [J]. ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY, 2022, 13 (04)
  • [2] FedDQ: Communication-Efficient Federated Learning with Descending Quantization
    Qu, Linping
    Song, Shenghui
    Tsui, Chi-Ying
    [J]. 2022 IEEE GLOBAL COMMUNICATIONS CONFERENCE (GLOBECOM 2022), 2022, : 281 - 286
  • [3] Communication-Efficient Agnostic Federated Averaging
    Ro, Jae
    Chen, Mingqing
    Mathews, Rajiv
    Mohri, Mehryar
    Suresh, Ananda Theertha
    [J]. INTERSPEECH 2021, 2021, : 871 - 875
  • [4] ADAPTIVE QUANTIZATION OF MODEL UPDATES FOR COMMUNICATION-EFFICIENT FEDERATED LEARNING
    Jhunjhunwala, Divyansh
    Gadhikar, Advait
    Joshi, Gauri
    Eldar, Yonina C.
    [J]. 2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021), 2021, : 3110 - 3114
  • [5] Communication-efficient federated learning
    Chen, Mingzhe
    Shlezinger, Nir
    Poor, H. Vincent
    Eldar, Yonina C.
    Cui, Shuguang
    [J]. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2021, 118 (17)
  • [6] DAdaQuant: Doubly-adaptive quantization for communication-efficient Federated Learning
    Hoenig, Robert
    Zhao, Yiren
    Mullins, Robert
    [J]. INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 162, 2022,
  • [7] A Communication-Efficient Federated Learning by Dynamic Quantization and Free-Ride Coding
    Chen, Junjie
    Wang, Qianfan
    Wan, Hai
    Ma, Xiao
    [J]. 2024 IEEE WIRELESS COMMUNICATIONS AND NETWORKING CONFERENCE, WCNC 2024, 2024,
  • [8] Communication-Efficient Vertical Federated Learning
    Khan, Afsana
    ten Thij, Marijn
    Wilbik, Anna
    [J]. ALGORITHMS, 2022, 15 (08)
  • [9] Communication-Efficient Adaptive Federated Learning
    Wang, Yujia
    Lin, Lu
    Chen, Jinghui
    [J]. INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 162, 2022,
  • [10] DEMYSTIFYING MODEL AVERAGING FOR COMMUNICATION-EFFICIENT FEDERATED MATRIX FACTORIZATION
    Wang, Shuai
    Suwandi, Richard Cornelius
    Chang, Tsung-Hui
    [J]. 2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021), 2021, : 3680 - 3684