How to Avoid Trivial Solutions in Physics-Informed Neural Networks

by   Raphael Leiteritz, et al.

The advent of scientific machine learning (SciML) has opened up a new field with many promises and challenges in the field of simulation science by developing approaches at the interface of physics- and data-based modelling. To this end, physics-informed neural networks (PINNs) have been introduced in recent years, which cope for the scarcity in training data by incorporating physics knowledge of the problem at so-called collocation points. In this work, we investigate the prediction performance of PINNs with respect to the number of collocation points used to enforce the physics-based penalty terms. We show that PINNs can fail, learning a trivial solution that fulfills the physics-derived penalty term by definition. We have developed an alternative sampling approach and a new penalty term enabling us to remedy this core problem of PINNs in data-scarce settings with competitive results while reducing the amount of collocation points needed by up to 80 % for benchmark problems.


Physics-informed neural networks for pathloss prediction

This paper introduces a physics-informed machine learning approach for p...

Good Lattice Training: Physics-Informed Neural Networks Accelerated by Number Theory

Physics-informed neural networks (PINNs) offer a novel and efficient app...

Eigenvalue initialisation and regularisation for Koopman autoencoders

Regularising the parameter matrices of neural networks is ubiquitous in ...

Physics-Informed Neural Networks for Shell Structures

The numerical modeling of thin shell structures is a challenge, which ha...

Learning Free-Surface Flow with Physics-Informed Neural Networks

The interface between data-driven learning methods and classical simulat...

Differentiable Physics-informed Graph Networks

While physics conveys knowledge of nature built from an interplay betwee...

Efficient Training of Physics-Informed Neural Networks with Direct Grid Refinement Algorithm

This research presents the development of an innovative algorithm tailor...

Please sign up or login with your details

Forgot password? Click here to reset