Paper "PDEBENCH: An Extensive Benchmark for Scientific Machine Learning" accepted at NeurIPS'22

November 11, 2022 /

Our Paper "PDEBENCH: An Extensive Benchmark for Scientific Machine Learning" got accepted at the NeurIPS 2022 Datasets & Benchmark Track

We are happy to announce that this years reviewers found our Paper "PDEBENCH: An Extensive Benchmark for Scientific Machine Learning" worthy of publishing at the NeurIPS 2022 Conference. In a group effort and close collaboration together with various other institutions, both from the University of Stuttgart and externally, we try to advance the current state of scientficially motivated benchmarks for machine learning. Our extensive suite contains 11 different scenarios of 1-, 2- and 3-D datasets generated from a wide range of PDE problems. Additionally we provide several implementations of baseline models with pre-trained weights to compare against. We hope this serves as a useful basis for many future developments in the space of Scientific Machine Learning.

To the top of the page