Interpretable conservation law estimation by deriving the symmetries of dynamics from trained deep neural networks

Y Mototake - Physical Review E, 2021 - APS
Understanding complex systems with their reduced model is one of the central roles in
scientific activities. Although physics has greatly been developed with the physical insights …

Exploiting Chaotic Dynamics as Deep Neural Networks

S Liu, N Akashi, Q Huang, Y Kuniyoshi… - arXiv preprint arXiv …, 2024 - arxiv.org
Chaos presents complex dynamics arising from nonlinearity and a sensitivity to initial states.
These characteristics suggest a depth of expressivity that underscores their potential for …

Signal identification without signal formulation

Y Mototake, Y Taguchi - arXiv preprint arXiv:2304.06522, 2023 - arxiv.org
When there are signals and noises, physicists try to identify signals by modeling them,
whereas statisticians oppositely try to model noise to identify signals. In this study, we …

Extracting Nonlinear Symmetries From Trained Neural Networks on Dynamics Data

Y Mototake - NeurIPS 2023 AI for Science Workshop - openreview.net
To support scientists who are developing the reduced model of complex physics systems,
we propose a method for extracting interpretable physics information from a deep neural …

[PDF][PDF] Conservation Law Estimation by Extracting the Symmetry of a Dynamical System Using a Deep Neural Network

Y Mototake - ml4physicalsciences.github.io
As deep neural networks (DNN) have the ability to model the distribution of datasets as a
low-dimensional manifold, we propose a method to extract the coordinate transformation …

[引用][C] On the Dynamics of Adversarial Input Attacks

Y Ji, T Wang

DNN を用いて抽出された力学系の対称性からの保存量推定

本武陽一 - 人工知能学会全国大会論文集第33 回(2019), 2019 - jstage.jst.go.jp
抄録 近年発達を続ける Deep Neural Networks (以下, DNN) が, 与えられたタスクを達成するため
に必要なデータセットの情報を, その分布を多様体としてモデル化することで 抽出する機能を持つこと …

[引用][C] DNNΛ༻ ͍ͯநग़͞Εͨྗֶܥͷରশੑ͔Βͷ อ ଘྔਪఆ