skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Learning continuous models for continuous physics
Dynamical systems that evolve continuously over time are ubiquitous throughout science and engineering. Machine learning (ML) provides data-driven approaches to model and predict the dynamics of such systems. A core issue with this approach is that ML models are typically trained on discrete data, using ML methodologies that are not aware of underlying continuity properties. This results in models that often do not capture any underlying continuous dynamics—either of the system of interest, or indeed of any related system. To address this challenge, we develop a convergence test based on numerical analysis theory. Our test verifies whether a model has learned a function that accurately approximates an underlying continuous dynamics. Models that fail this test fail to capture relevant dynamics, rendering them of limited utility for many scientific prediction tasks; while models that pass this test enable both better interpolation and better extrapolation in multiple ways. Our results illustrate how principled numerical analysis methods can be coupled with existing ML training/testing methodologies to validate models for science and engineering applications.  more » « less
Award ID(s):
2319621
PAR ID:
10519656
Author(s) / Creator(s):
; ; ;
Publisher / Repository:
Springer
Date Published:
Journal Name:
Communications Physics
Volume:
6
Issue:
1
ISSN:
2399-3650
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; (, Transactions on machine learning research)
    Identifying hidden interactions within complex systems is key to unlocking deeper insights into their operational dynamics, including how their elements affect each other and contribute to the overall system behavior. For instance, in neuroscience, discovering neuron-to-neuron interactions is essential for understanding brain function; in ecology, recognizing interactions among populations is key to understanding complex ecosystems. Such systems, often modeled as dynamical systems, typically exhibit noisy high-dimensional and non-stationary temporal behavior that renders their identification challenging. Existing dynamical system identification methods typically yield operators that accurately capture short-term behavior but fail to predict long-term trends, suggesting an incomplete capture of the underlying process. Methods that consider extended forecasts (e.g., recurrent neural networks) lack explicit representations of element-wise interactions and require substantial training data, thereby failing to capture interpretable network operators. Here we introduce Lookahead-driven Inference of Networked Operators for Continuous Stability (LINOCS), a robust learning procedure for identifying hidden dynamical interactions in noisy time-series data. LINOCS integrates several multi-step predictions with adaptive weights during training to recover dynamical operators that can yield accurate long-term predictions. We demonstrate LINOCS’ ability to recover the ground truth dynamical operators underlying synthetic time-series data for multiple dynamical systems models (including linear, piece-wise linear, time-changing linear systems’ decomposition, and regularized linear time-varying systems) as well as its capability to produce meaningful operators with robust reconstructions through various real-world examples 
    more » « less
  2. ; (, Geo-Congress 2023)
    Rathje, E.; Montoya, B.; Wayne, M. (Ed.)
    The rise of data capture and storage capabilities have led to greater data granularity and sharing of data sets in geotechnical earthquake engineering. This broader shift to big data requires ways to process and extract value from it and is aided by the progress in methodologies from the computer science domain and advancements in computer hardware capabilities. General machine learning (ML) models typically receive a set of input parameters and run them through an algorithm to gain outputs with no constraints on the parameters or algorithm process. Three topic areas of ML applications in geotechnical earthquake engineering are reviewed and summarized in this paper: seismic response, liquefaction triggering analysis, and performance-based assessments (lateral displacements and settlement analysis). The current progress of ML is summarized, while the challenges and potential in adopting such approaches are addressed. 
    more » « less
  3. ; ; ; ; (, ACM Transactions on Intelligent Systems and Technology)
    Urban traffic status (e.g., traffic speed and volume) is highly dynamic in nature, namely, varying across space and evolving over time. Thus, predicting such traffic dynamics is of great importance to urban development and transportation management. However, it is very challenging to solve this problem due to spatial-temporal dependencies and traffic uncertainties. In this article, we solve the traffic dynamics prediction problem from Bayesian meta-learning perspective and propose a novel continuous spatial-temporal meta-learner (cST-ML), which is trained on a distribution of traffic prediction tasks segmented by historical traffic data with the goal of learning a strategy that can be quickly adapted to related but unseen traffic prediction tasks. cST-ML tackles the traffic dynamics prediction challenges by advancing the Bayesian black-box meta-learning framework through the following new points: (1) cST-ML captures the dynamics of traffic prediction tasks using variational inference, and to better capture the temporal uncertainties within tasks, cST-ML performs as a rolling window within each task; (2) cST-ML has novel designs in architecture, where CNN and LSTM are embedded to capture the spatial-temporal dependencies between traffic status and traffic-related features; (3) novel training and testing algorithms for cST-ML are designed. We also conduct experiments on two real-world traffic datasets (taxi inflow and traffic speed) to evaluate our proposed cST-ML. The experimental results verify that cST-ML can significantly improve the urban traffic prediction performance and outperform all baseline models especially when obvious traffic dynamics and temporal uncertainties are presented. 
    more » « less
  4. ; (, 60th IEEE Conference on Decision and Control)
    This paper offers new analytical conditions on the system parameters of a particular class of planar dynamical systems which would allow them to undergo a Hopf bifurcation. These systems are constructed as a means of generating multiple behaviors from the same single continuous dynamical system model, without resorting to switching between distinct component continuous dynamics associated to each behavioral mode. This work builds on recent advances which introduced motivation dynamics as an efficient way to design multi-behavioral systems. The contribution of this paper is that it expands the scope of the motivation dynamics approach, and offers explicit analytic conditions on the system parameters to guarantee the existence of bifurcations, which can then be utilized to better engineer the structure and location of the resulting equilibria. Numerical simulations confirm the theoretical predictions for the onset of the Hopf bifurcations. 
    more » « less
  5. ; ; ; (, International Conference Information Visualisation (IV))
    Kinematic motion analysis is widely used in health-care, sports medicine, robotics, biomechanics, sports science, etc. Motion capture systems are essential for motion analysis. There are three types of motion capture systems: marker-based capture, vision-based capture, and volumetric capture. Marker-based motion capture systems can achieve fairly accurate results but attaching markers to a body is inconvenient and time-consuming. Vision-based, marker-less motion capture systems are more desirable because of their non-intrusiveness and flexibility. Volumetric capture is a newer and more advanced marker-less motion capture system that can reconstruct realistic, full-body, animated 3D character models. But volumetric capture has rarely been used for motion analysis because volumetric motion data presents new challenges. We propose a new method for conducting kinematic motion analysis using volumetric capture data. This method consists of a three-stage pipeline. First, the motion is captured by a volumetric capture system. Then the volumetric capture data is processed using the Iterative Closest Points (ICP) algorithm to generate virtual markers that track the motion. Third, the motion tracking data is imported into the biomechanical analysis tool OpenSim for kinematic motion analysis. Our motion analysis method enables users to apply numerical motion analysis to the skeleton model in OpenSim while also studying the full-body, animated 3D model from different angles. It has the potential to provide more detailed and in-depth motion analysis for areas such as healthcare, sports science, and biomechanics. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email