On instabilities in neural network-based physics simulators

• URL: http://arxiv.org/abs/2406.13101v1
• Date: Tue, 18 Jun 2024 23:25:14 GMT
• Title: On instabilities in neural network-based physics simulators
• Authors: Daniel Floryan,
• Abstract summary: Long-time dynamics produced by neural networks are often unphysical or unstable. We show that the rate of convergence of the training dynamics is uneven and depends on the distribution of energy in the data. Injecting synthetic noise into the data during training adds damping to the training dynamics and can stabilize the learned simulator.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: When neural networks are trained from data to simulate the dynamics of physical systems, they encounter a persistent challenge: the long-time dynamics they produce are often unphysical or unstable. We analyze the origin of such instabilities when learning linear dynamical systems, focusing on the training dynamics. We make several analytical findings which empirical observations suggest extend to nonlinear dynamical systems. First, the rate of convergence of the training dynamics is uneven and depends on the distribution of energy in the data. As a special case, the dynamics in directions where the data have no energy cannot be learned. Second, in the unlearnable directions, the dynamics produced by the neural network depend on the weight initialization, and common weight initialization schemes can produce unstable dynamics. Third, injecting synthetic noise into the data during training adds damping to the training dynamics and can stabilize the learned simulator, though doing so undesirably biases the learned dynamics. For each contributor to instability, we suggest mitigative strategies. We also highlight important differences between learning discrete-time and continuous-time dynamics, and discuss extensions to nonlinear systems.

Related papers

• Learning System Dynamics without Forgetting [60.08612207170659]
  Predicting trajectories of systems with unknown dynamics is crucial in various research fields, including physics and biology. We present a novel framework of Mode-switching Graph ODE (MS-GODE), which can continually learn varying dynamics. We construct a novel benchmark of biological dynamic systems, featuring diverse systems with disparate dynamics.
  arXiv  Detail & Related papers  (2024-06-30T14:55:18Z)
• Learning Dissipative Neural Dynamical Systems [0.8993153817914281]
  In general, imposing dissipativity constraints during neural network training is a hard problem for which no known techniques exist. We show that these two perturbation problems can be solved independently to obtain a neural dynamical model guaranteed to be dissipative.
  arXiv  Detail & Related papers  (2023-09-27T21:25:26Z)
• Leveraging Neural Koopman Operators to Learn Continuous Representations of Dynamical Systems from Scarce Data [0.0]
  We propose a new deep Koopman framework that represents dynamics in an intrinsically continuous way. This framework leads to better performance on limited training data.
  arXiv  Detail & Related papers  (2023-03-13T10:16:19Z)
• Critical Learning Periods for Multisensory Integration in Deep Networks [112.40005682521638]
  We show that the ability of a neural network to integrate information from diverse sources hinges critically on being exposed to properly correlated signals during the early phases of training. We show that critical periods arise from the complex and unstable early transient dynamics, which are decisive of final performance of the trained system and their learned representations.
  arXiv  Detail & Related papers  (2022-10-06T23:50:38Z)
• Decomposed Linear Dynamical Systems (dLDS) for learning the latent components of neural dynamics [6.829711787905569]
  We propose a new decomposed dynamical system model that represents complex non-stationary and nonlinear dynamics of time series data. Our model is trained through a dictionary learning procedure, where we leverage recent results in tracking sparse vectors over time. In both continuous-time and discrete-time instructional examples we demonstrate that our model can well approximate the original system.
  arXiv  Detail & Related papers  (2022-06-07T02:25:38Z)
• Learning Fine Scale Dynamics from Coarse Observations via Inner Recurrence [0.0]
  Recent work has focused on data-driven learning of the evolution of unknown systems via deep neural networks (DNNs) This paper presents a computational technique to learn the fine-scale dynamics from such coarsely observed data.
  arXiv  Detail & Related papers  (2022-06-03T20:28:52Z)
• Capturing Actionable Dynamics with Structured Latent Ordinary Differential Equations [68.62843292346813]
  We propose a structured latent ODE model that captures system input variations within its latent representation. Building on a static variable specification, our model learns factors of variation for each input to the system, thus separating the effects of the system inputs in the latent space.
  arXiv  Detail & Related papers  (2022-02-25T20:00:56Z)
• Constructing Neural Network-Based Models for Simulating Dynamical Systems [59.0861954179401]
  Data-driven modeling is an alternative paradigm that seeks to learn an approximation of the dynamics of a system using observations of the true system. This paper provides a survey of the different ways to construct models of dynamical systems using neural networks. In addition to the basic overview, we review the related literature and outline the most significant challenges from numerical simulations that this modeling paradigm must overcome.
  arXiv  Detail & Related papers  (2021-11-02T10:51:42Z)
• Learning Contact Dynamics using Physically Structured Neural Networks [81.73947303886753]
  We use connections between deep neural networks and differential equations to design a family of deep network architectures for representing contact dynamics between objects. We show that these networks can learn discontinuous contact events in a data-efficient manner from noisy observations. Our results indicate that an idealised form of touch feedback is a key component of making this learning problem tractable.
  arXiv  Detail & Related papers  (2021-02-22T17:33:51Z)
• Physics-Incorporated Convolutional Recurrent Neural Networks for Source Identification and Forecasting of Dynamical Systems [10.689157154434499]
  In this paper, we present a hybrid framework combining numerical physics-based models with deep learning for source identification. We formulate our model PhICNet as a convolutional recurrent neural network (RNN) which is end-to-end trainable for predicting S-temporal evolution. Experimental results show that the proposed model can forecast the dynamics for a relatively long time and identify the sources as well.
  arXiv  Detail & Related papers  (2020-04-14T00:27:18Z)
• Learning Stable Deep Dynamics Models [91.90131512825504]
  We propose an approach for learning dynamical systems that are guaranteed to be stable over the entire state space. We show that such learning systems are able to model simple dynamical systems and can be combined with additional deep generative models to learn complex dynamics.
  arXiv  Detail & Related papers  (2020-01-17T00:04:45Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.