PINNs

We introduce geometry-informed neural networks (GINNs) to train shape generative models without any data.

We present how to use Lie Point Symmetries of PDEs to improve physics-informed neural networks. Published at NeurIPS 2023.