Phil Sidney Ostheimer — accepted AI-conference papers

2026

Heavy-tailed Physics-Informed Neural Networks

Jephte Abijuru, Mayank Kumar Nagda, Phil Sidney Ostheimer, Jan Tauberschmidt, Sebastian Vollmer, Stephan Mandt +2

ICML 2026poster

Physics-informed neural networks (PINNs) enforce physical laws by minimizing partial differential equation (PDE) residuals and auxiliary constraints. Standard training relies on a mean-squared error (MSE) objective, which implicitly assumes independent Gaussian residuals with a fixed global variance…

Cited by 0 Source ↗Scholar ↗

2026

Physics-Informed Residual Flows

Jephte Abijuru, Mayank Kumar Nagda, Phil Sidney Ostheimer, Sebastian Vollmer, Marius Kloft, Sophie Fellenz

ICML 2026poster

Physics-Informed Neural Networks (PINNs) embed physical laws into deep learning models. However, conventional PINNs often suffer from failure modes leading to inaccurate solutions. We trace these failure modes to two structural pathologies: gradient shattering, where gradients degrade with depth and…

Cited by 0 Source ↗Scholar ↗

2026

Skipping the Zeros in Diffusion Models for Sparse Data Generation

Phil Sidney Ostheimer, Mayank Kumar Nagda, Andriy Balinskyy, Gabriel Rodrigues, Jean Radig, Carl Herrmann +3

ICML 2026poster

Diffusion models (DMs) excel on dense continuous data, but are not designed for sparse continuous data. They do not model exact zeros that represent the deliberate absence of a signal. As a result, they erase sparsity patterns and perform unnecessary computation on mostly zero entries. With Sparsity…

Cited by 0 Source ↗Scholar ↗