Abstract
We study first order methods to compute the barycenter of a probability distribution P over the space of probability measures with finite second moment. We develop a framework to derive global rates of convergence for both gradient descent and stochastic gradient descent despite the fact that the barycenter functional is not geodesically convex. Our analysis overcomes this technical hurdle by employing a Polyak-Łojasiewicz (PL) inequality and relies on tools from optimal transport and metric geometry. In turn, we establish a PL inequality when P is supported on the Bures-Wasserstein manifold of Gaussian probability measures. It leads to the first global rates of convergence for first order methods in this context.
| Original language | English |
|---|---|
| Pages (from-to) | 1276-1304 |
| Number of pages | 29 |
| Journal | Proceedings of Machine Learning Research |
| Volume | 125 |
| Publication status | Published - 1 Jan 2020 |
| Externally published | Yes |
| Event | 33rd Conference on Learning Theory, COLT 2020 - Virtual, Online, Austria Duration: 9 Jul 2020 → 12 Jul 2020 |
Keywords
- Wasserstein barycenters
- geodesic optimization
- optimal transport
Fingerprint
Dive into the research topics of 'Gradient descent algorithms for Bures-Wasserstein barycenters'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver