Home Weather Adapting Noise to Data: Generative Flows from 1D Processes
Weather

Adapting Noise to Data: Generative Flows from 1D Processes

Key Points

arXiv:2510.12636v5 Announce Type: replace-cross Abstract: The default Gaussian latent in flow-based generative models poses challenges when learning certain distributions such as heavy-tailed ones. We introduce a general framework for learning data-adaptive parametric prior distributions (latent noise) using one-dimensional quantile functions, optimized via the Wasserstein distance between noise and data. The quantile-based prior parameterization naturally adapts to both heavy-tailed and...

arXiv:2510.12636v5 Announce Type: replace-cross Abstract: The default Gaussian latent in flow-based generative models poses challenges when learning certain distributions such as heavy-tailed ones. We introduce a general framework for learning data-adaptive parametric prior distributions (latent noise) using one-dimensional quantile functions, optimized via the Wasserstein distance between noise and data. The quantile-based prior parameterization naturally adapts to both heavy-tailed and compactly supported distributions and shortens transport paths. Numerical results on heavy-tailed weather and image datasets confirm the method's flexibility and effectiveness achieved with negligible computational overhead.
Adapting Noise (ORG) Wasserstein (PERSON)
Originally published by arXiv CS Read original →