Rafael Bischof – Medium

Rafael Bischof
in
Towards Data Science

4 Ideas for Physics-Informed Neural Networks that failed

Here is a list of extensions for PINNs that either did not improve their performance, or broke them completely — so you do not have to try…

9 min readFeb 11, 2023

--

2

4 Ideas for Physics-Informed Neural Networks that failed

--

2

Rafael Bischof
in
Towards Data Science

10 Useful Hints and Tricks for Improving Physics-Informed Neural Networks (PINNs)

A handful of tricks concerning the sampling, architecture, activation function, loss balancing, optimisers, data normalisation, and more.

10 min readFeb 8, 2023

--

2

10 Useful Hints and Tricks for Improving Physics-Informed Neural Networks (PINNs)

--

2

Rafael Bischof

AI-Augmented Designing Process of a Pedestrian Bridge in Switzerland

Leveraging AI to enhance designing processes in the AEC Industry: A Case Study of a Pedestrian Bridge in Switzerland

7 min readFeb 6, 2023

--

AI-Augmented Designing Process of a Pedestrian Bridge in Switzerland

--

Rafael Bischof
in
Towards Data Science

Mixture of Experts for PINNs (MoE-PINNs)

Utilising Ensembles to enhance Physics-Informed Neural Networks

9 min readFeb 2, 2023

--

1

Mixture of Experts for PINNs (MoE-PINNs)

--

1

Rafael Bischof
in
Towards Data Science

Improving Physics-Informed Neural Networks through Adaptive Loss Balancing

How to boost your PINN’s performance with ReLoBRaLo, Learning Rate Annealing and co.

14 min readJan 31, 2023

--

2

Improving Physics-Informed Neural Networks through Adaptive Loss Balancing

--

2

Rafael Bischof

Rafael Bischof

My research focuses on leveraging machine learning to solve engineering challenges and incorporating prior domain knowledge to improve performance..

Following

Help
Status
About
Careers
Press
Blog
Privacy
Terms
Text to speech
Teams