A computational topology library

Areas: Data and visualization, Artificial intelligence

CASS member: RAPIDS

Description

Oineus is a computational topology library; it can compute a topological descriptor that captures the shape of data at different scales and provides different ways to optimize it. For PyTorch users, the descriptor is provided simply as a differentiable tensor and all internal details of the optimization are handled by the library. There is an alternative for Jax/Tensorflow users which is slightly more verbose.

Target audience

Machine learning researchers who are interested in topological data analysis methods.

License: BSD-3-Clause

Additional resources