Learning to Optimize with Dynamic Mode Decomposition

被引:0
|
作者
Simanek, Petr [1 ,2 ]
Vasata, Daniel [1 ]
Kordik, Pavel [1 ]
机构
[1] Czech Tech Univ, Fac Informat Technol, Prague, Czech Republic
[2] GoodAI, Prague, Czech Republic
关键词
machine learning; recurrent neural networks; optimization;
D O I
10.1109/IJCNN55064.2022.9892364
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Designing faster optimization algorithms is of ever-growing interest. In recent years, learning to learn methods that learn how to optimize demonstrated very encouraging results. Current approaches usually do not effectively include the dynamics of the optimization process during training. They either omit it entirely or only implicitly assume the dynamics of an isolated parameter. In this paper, we show how to utilize the dynamic mode decomposition method for extracting informative features about optimization dynamics. By employing those features, we show that our learned optimizer generalizes much better to unseen optimization problems in short. The improved generalization is illustrated on multiple tasks where training the optimizer on one neural network generalizes to different architectures and distinct datasets.
引用
收藏
页数:8
相关论文
共 50 条
  • [41] PREDICTION ACCURACY OF DYNAMIC MODE DECOMPOSITION
    Lu, Hannah
    Tartakovsky, Daniel M.
    [J]. SIAM JOURNAL ON SCIENTIFIC COMPUTING, 2020, 42 (03): : A1639 - A1662
  • [42] SPARSE NONNEGATIVE DYNAMIC MODE DECOMPOSITION
    Takeishi, Naoya
    Kawahara, Yoshinobu
    Yairi, Takehisa
    [J]. 2017 24TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2017, : 2682 - 2686
  • [43] On the convergence of Hermitian Dynamic Mode Decomposition
    Boullé, Nicolas
    Colbrook, Matthew J.
    [J]. Physica D: Nonlinear Phenomena, 2025, 472
  • [44] EVALUATING THE ACCURACY OF THE DYNAMIC MODE DECOMPOSITION
    Zhang, Hao
    Dawson, Scott T. M.
    Rowley, Clarence W.
    Deem, Eric A.
    Cattafesta, Louis N.
    [J]. JOURNAL OF COMPUTATIONAL DYNAMICS, 2020, 7 (01): : 35 - 56
  • [45] Towards an Adaptive Dynamic Mode Decomposition
    Murshed, Mohammad N.
    Uddin, M. Monir
    [J]. RESULTS IN CONTROL AND OPTIMIZATION, 2022, 6
  • [46] Extended dynamic mode decomposition with dictionary learning: A data-driven adaptive spectral decomposition of the Koopman operator
    Li, Qianxiao
    Dietrich, Felix
    Bollt, Erik M.
    Kevrekidis, Ioannis G.
    [J]. CHAOS, 2017, 27 (10)
  • [47] Identification of dynamic textures using Dynamic Mode Decomposition
    Previtali, D.
    Valceschini, N.
    Mazzoleni, M.
    Previdi, F.
    [J]. IFAC PAPERSONLINE, 2020, 53 (02): : 2423 - 2428
  • [48] Extended dynamic mode decomposition with dictionary learning using neural ordinary differential equations
    Terao, Hiroaki
    Shirasaka, Sho
    Suzuki, Hideyuki
    [J]. IEICE NONLINEAR THEORY AND ITS APPLICATIONS, 2021, 12 (04): : 626 - 638
  • [49] Analysis of Dynamic Stall Using Dynamic Mode Decomposition Technique
    Mariappan, Sathesh
    Gardner, A. D.
    Richter, Kai
    Raffel, Markus
    [J]. AIAA JOURNAL, 2014, 52 (11) : 2427 - 2439
  • [50] Uncertainty analysis of dynamic mode decomposition for xenon dynamic forecasting
    Liu, Jianpeng
    Gong, Helin
    Wang, Zhiyong
    Li, Qing
    [J]. ANNALS OF NUCLEAR ENERGY, 2023, 194