Practical and Fast Momentum-Based Power Methods

被引:0
|
作者
Rabbani, Tahseen [1 ]
Jain, Apollo [2 ]
Rajkumar, Arjun [1 ]
Huang, Furong [1 ]
机构
[1] Univ Maryland, Dept Comp Sci, College Pk, MD 20742 USA
[2] Syst & Technol Res, Sensors Div, Arlington, VA USA
基金
美国国家科学基金会;
关键词
matrix decomposition; PCA; power methods; momentum acceleration; streaming PCA;
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
The power method is a classical algorithm with broad applications in machine learning tasks, including streaming PCA, spectral clustering, and low-rank matrix approximation. The distilled purpose of the vanilla power method is to determine the largest eigenvalue (in absolute modulus) and its eigenvector of a matrix. A momentum-based scheme can be used to accelerate the power method, but achieving an optimal convergence rate with existing algorithms critically relies on additional spectral information that is unavailable at run-time, and sub-optimal initializations can result in divergence. In this paper, we provide a pair of novel momentum-based power methods, which we call the delayed momentum power method (DMPower) and a streaming variant, the delayed momentum streaming method (DMStream). Our methods leverage inexact deflation and are capable of achieving near-optimal convergence with far less restrictive hyperparameter requirements. We provide convergence analyses for both algorithms through the lens of perturbation theory. Further, we experimentally demonstrate that DMPower routinely outperforms the vanilla power method and that both algorithms match the convergence speed of an oracle running existing accelerated methods with perfect spectral knowledge.
引用
收藏
页码:721 / 756
页数:36
相关论文
共 50 条
  • [1] Momentum-Based Policy Gradient Methods
    Huang, Feihu
    Gao, Shangqian
    Pei, Jian
    Huang, Heng
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 119, 2020, 119
  • [2] GENERALIZED MOMENTUM-BASED METHODS: A HAMILTONIAN PERSPECTIVE
    Diakonikolas, Jelena
    Jordan, Michael, I
    SIAM JOURNAL ON OPTIMIZATION, 2021, 31 (01) : 915 - 944
  • [3] Fast-MoCo: Boost Momentum-Based Contrastive Learning with Combinatorial Patches
    Ci, Yuanzheng
    Lin, Chen
    Bai, Lei
    Ouyang, Wanli
    COMPUTER VISION, ECCV 2022, PT XXVI, 2022, 13686 : 290 - 306
  • [4] Momentum-based wavelet and double wavelet neural networks for power system applications
    Subramaniam Nachimuthu Deepa
    John Basha Rizwana
    Neural Computing and Applications, 2018, 29 : 495 - 511
  • [5] Momentum-based wavelet and double wavelet neural networks for power system applications
    Deepa, Subramaniam Nachimuthu
    Rizwana, John Basha
    NEURAL COMPUTING & APPLICATIONS, 2018, 29 (07): : 495 - 511
  • [6] Noise amplifiation of momentum-based optimization algorithms
    Mohammadi, Hesameddin
    Razaviyayn, Meisam
    Jovanovic, Mihailo R.
    2023 AMERICAN CONTROL CONFERENCE, ACC, 2023, : 849 - 854
  • [7] A momentum-based deformation system for granular material
    Zeng, Ya-Lun
    Tan, Charlie Irawan
    Tai, Wen-Kai
    Yang, Mau-Tsuen
    Chiang, Cheng-Chin
    Chang, Chin-Chen
    COMPUTER ANIMATION AND VIRTUAL WORLDS, 2007, 18 (4-5) : 289 - 300
  • [8] Superconductivity induced by fluctuations of momentum-based multipoles
    Sumita, Shuntaro
    Yanase, Youichi
    PHYSICAL REVIEW RESEARCH, 2020, 2 (03):
  • [9] Momentum-Based Contextual Federated Reinforcement Learning
    Yue, Sheng
    Hua, Xingyuan
    Deng, Yongheng
    Chen, Lili
    Ren, Ju
    Zhang, Yaoxue
    IEEE-ACM TRANSACTIONS ON NETWORKING, 2024,
  • [10] Momentum-Based Adaptive Laws for Identification and Control
    Somers, Luke
    Haddad, Wassim M.
    AEROSPACE, 2024, 11 (12)