
A PhysicsInformed Deep Learning Paradigm for CarFollowing Models
Carfollowing behavior has been extensively studied using physicsbased ...
read it

Mutual Information for Explainable Deep Learning of Multiscale Systems
Timely completion of design cycles for multiscale and multiphysics syste...
read it

Differentiable Physics Models for Realworld Offline Modelbased Reinforcement Learning
A limitation of modelbased reinforcement learning (MBRL) is the exploit...
read it

An energybased error bound of physicsinformed neural network solutions in elasticity
An energybased a posteriori error bound is proposed for the physicsinf...
read it

Physicsinformed linear regression is a competitive approach compared to Machine Learning methods in building MPC
Because physicsbased building models are difficult to obtain as each bu...
read it

Learning and Optimization with Bayesian Hybrid Models
Bayesian hybrid models fuse physicsbased insights with machine learning...
read it

PhysicsInformed Neural Networks (PINNs) for Sound Field Predictions with Parameterized Sources and Impedance Boundaries
Realistic sound is essential in virtual environments, such as computer g...
read it
GINNs: GraphInformed Neural Networks for Multiscale Physics
We introduce the concept of a GraphInformed Neural Network (GINN), a hybrid approach combining deep learning with probabilistic graphical models (PGMs) that acts as a surrogate for physicsbased representations of multiscale and multiphysics systems. GINNs address the twin challenges of removing intrinsic computational bottlenecks in physicsbased models and generating large data sets for estimating probability distributions of quantities of interest (QoIs) with a high degree of confidence. Both the selection of the complex physics learned by the NN and its supervised learning/prediction are informed by the PGM, which includes the formulation of structured priors for tunable control variables (CVs) to account for their mutual correlations and ensure physically sound CV and QoI distributions. GINNs accelerate the prediction of QoIs essential for simulationbased decisionmaking where generating sufficient sample data using physicsbased models alone is often prohibitively expensive. Using a realworld application grounded in supercapacitorbased energy storage, we describe the construction of GINNs from a Bayesian networkembedded homogenized model for supercapacitor dynamics, and demonstrate their ability to produce kernel density estimates of relevant nonGaussian, skewed QoIs with tight confidence intervals.
READ FULL TEXT
Comments
There are no comments yet.