Scale-invariant Learning by Physics Inversion

被引:0
|
作者
Holl, Philipp [1 ]
Koltun, Vladlen [2 ]
Thuerey, Nils [1 ]
机构
[1] Tech Univ Munich, Munich, Germany
[2] Apple, Cupertino, CA USA
来源
ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35 (NEURIPS 2022) | 2022年
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
Solving inverse problems, such as parameter estimation and optimal control, is a vital part of science. Many experiments repeatedly collect data and rely on machine learning algorithms to quickly infer solutions to the associated inverse problems. We find that state-of-the-art training techniques are not well-suited to many problems that involve physical processes. The highly nonlinear behavior, common in physical processes, results in strongly varying gradients that lead first-order optimizers like SGD or Adam to compute suboptimal optimization directions. We propose a novel hybrid training approach that combines higher-order optimization methods with machine learning techniques. We take updates from a scale-invariant inverse problem solver and embed them into the gradient-descent-based learning pipeline, replacing the regular gradient of the physical process. We demonstrate the capabilities of our method on a variety of canonical physical systems, showing that it yields significant improvements on a wide range of optimization and learning problems.
引用
收藏
页数:14
相关论文
共 50 条
  • [31] SCALE-INVARIANT EXTENDED INFLATION
    HOLMAN, R
    KOLB, EW
    VADAS, SL
    WANG, Y
    PHYSICAL REVIEW D, 1991, 43 (12): : 3833 - 3845
  • [32] Scale-invariant gravity: geometrodynamics
    Anderson, E
    Barbour, J
    Foster, B
    O Murchadha, N
    CLASSICAL AND QUANTUM GRAVITY, 2003, 20 (08) : 1571 - 1604
  • [33] SCALE-INVARIANT CORNER KEYPOINTS
    Li, Bo
    Li, Haibo
    Soderstrom, Ulrik
    2014 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2014, : 5741 - 5745
  • [34] Dominance and Optimisation Based on Scale-Invariant Maximum Margin Preference Learning
    Montazery, Mojtaba
    Wilson, Nic
    PROCEEDINGS OF THE TWENTY-SIXTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2017, : 1209 - 1215
  • [35] Scale-Invariant Reinforcement Learning in Real-Time Strategy Games
    Diniz Lemos, Marcelo Luiz Harry
    Vieira, Ronaldo e Silva
    Rocha Tavares, Anderson
    Soriano Marcolino, Leandro
    Chaimowicz, Luiz
    PROCEEDINGS OF THE 22ND BRAZILIAN SYMPOSIUM ON COMPUTER GAMES AND DIGITAL ENTERTAINMENT, SBGAMES, 2023, 2023, : 11 - 19
  • [36] Scale-invariant resonance tagging in multijet events and new physics in Higgs pair production
    Maxime Gouzevitch
    Alexandra Oliveira
    Juan Rojo
    Rogerio Rosenfeld
    Gavin P. Salam
    Veronica Sanz
    Journal of High Energy Physics, 2013
  • [37] Scale-invariant resonance tagging in multijet events and new physics in Higgs pair production
    Gouzevitch, Maxime
    Oliveira, Alexandra
    Rojo, Juan
    Rosenfeld, Rogerio
    Salam, Gavin P.
    Sanz, Veronica
    JOURNAL OF HIGH ENERGY PHYSICS, 2013, (07):
  • [38] Learning How to Extract Rotation-Invariant and Scale-Invariant Features from Texture Images
    Javier A. Montoya-Zegarra
    João Paulo Papa
    Neucimar J. Leite
    Ricardo da Silva Torres
    Alexandre Falcão
    EURASIP Journal on Advances in Signal Processing, 2008
  • [39] Learning how to extract rotation-invariant and scale-invariant features from texture images
    Montoya-Zegarra, Javier A.
    Paulo Papa, Joao
    Leite, Neucimar J.
    da Silva Torres, Ricardo
    Falcao, Alexandre X.
    EURASIP JOURNAL ON ADVANCES IN SIGNAL PROCESSING, 2008, 2008 (1)
  • [40] Scale-invariant breaking of conformal symmetry
    Dymarsky, Anatoly
    Zhiboedov, Alexander
    JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL, 2015, 48 (41)