FedNS: A Fast Sketching Newton-Type Algorithm for Federated Learning

被引:0
|
作者
Li, Jian [1 ]
Liu, Yong [2 ]
Wang, Weiping [1 ]
机构
[1] Chinese Acad Sci, Inst Informat Engn, Beijing, Peoples R China
[2] Renmin Univ China, Gaoling Sch Artificial Intelligence, Beijing, Peoples R China
基金
中国博士后科学基金; 中国国家自然科学基金;
关键词
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
Recent Newton-type federated learning algorithms have demonstrated linear convergence with respect to the communication rounds. However, communicating Hessian matrices is often unfeasible due to their quadratic communication complexity. In this paper, we introduce a novel approach to tackle this issue while still achieving fast convergence rates. Our proposed method, named as Federated Newton Sketch methods (FedNS), approximates the centralized Newton's method by communicating the sketched square-root Hessian instead of the exact Hessian. To enhance communication efficiency, we reduce the sketch size to match the effective dimension of the Hessian matrix. We provide convergence analysis based on statistical learning for the federated Newton sketch approaches. Specifically, our approaches reach super-linear convergence rates w.r.t. the communication rounds for the first time. We validate the effectiveness of our algorithms through various experiments, which coincide with our theoretical findings.
引用
收藏
页码:13509 / 13517
页数:9
相关论文
共 50 条
  • [21] On the DSM Newton-type method
    Ramm, A. G.
    JOURNAL OF APPLIED MATHEMATICS AND COMPUTING, 2012, 38 (1-2) : 523 - 533
  • [22] Communication-Efficient Federated Learning: A Second Order Newton-Type Method With Analog Over-the-Air Aggregation
    Krouka, Mounssif
    Elgabli, Anis
    Ben Issaid, Chaouki
    Bennis, Mehdi
    IEEE TRANSACTIONS ON GREEN COMMUNICATIONS AND NETWORKING, 2022, 6 (03): : 1862 - 1874
  • [23] Inexact Newton-type methods
    Argyros, Ioannis K.
    Hilout, Said
    JOURNAL OF COMPLEXITY, 2010, 26 (06) : 577 - 590
  • [24] On the DSM Newton-type method
    A. G. Ramm
    Journal of Applied Mathematics and Computing, 2012, 38 (1-2) : 523 - 533
  • [25] Improved recursive Newton-type algorithm for power system relaying and measurement
    Terzija, VV
    IEE PROCEEDINGS-GENERATION TRANSMISSION AND DISTRIBUTION, 1998, 145 (01) : 15 - 20
  • [26] A smoothing projected Newton-type algorithm for semi-infinite programming
    Qi, Liqun
    Ling, Chen
    Tong, Xiaojiao
    Zhou, Guanglu
    COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2009, 42 (01) : 1 - 30
  • [27] A regularized smoothing Newton-type algorithm for quasi-variational inequalities
    Ni, Tie
    Zhai, Jun
    COMPUTERS & MATHEMATICS WITH APPLICATIONS, 2014, 68 (10) : 1312 - 1324
  • [28] Newton-type algorithm for the control of power flow in electrical power networks
    FuerteEsquivel, CR
    Acha, E
    IEEE TRANSACTIONS ON POWER SYSTEMS, 1997, 12 (04) : 1474 - 1480
  • [29] SVC and TCSC implemented into a Newton-Type Harmonic Power Flow algorithm
    Chavez, J. J.
    Ramirez, A.
    2006 38TH ANNUAL NORTH AMERICAN POWER SYMPOSIUM, NAPS-2006 PROCEEDINGS, 2006, : 283 - +
  • [30] A smoothing projected Newton-type algorithm for semi-infinite programming
    Liqun Qi
    Chen Ling
    Xiaojiao Tong
    Guanglu Zhou
    Computational Optimization and Applications, 2009, 42 : 1 - 30