laplax -- Laplace Approximations with JAX
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913955513892864 |
|---|---|
| author | Weber, Tobias Mucsányi, Bálint Rommel, Lenard Christie, Thomas Kasüschke, Lars Pförtner, Marvin Hennig, Philipp |
| author_facet | Weber, Tobias Mucsányi, Bálint Rommel, Lenard Christie, Thomas Kasüschke, Lars Pförtner, Marvin Hennig, Philipp |
| contents | The Laplace approximation provides a scalable and efficient means of quantifying weight-space uncertainty in deep neural networks, enabling the application of Bayesian tools such as predictive uncertainty and model selection via Occam's razor. In this work, we introduce laplax, a new open-source Python package for performing Laplace approximations with jax. Designed with a modular and purely functional architecture and minimal external dependencies, laplax offers a flexible and researcher-friendly framework for rapid prototyping and experimentation. Its goal is to facilitate research on Bayesian neural networks, uncertainty quantification for deep learning, and the development of improved Laplace approximation techniques. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_17013 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | laplax -- Laplace Approximations with JAX Weber, Tobias Mucsányi, Bálint Rommel, Lenard Christie, Thomas Kasüschke, Lars Pförtner, Marvin Hennig, Philipp Machine Learning Artificial Intelligence The Laplace approximation provides a scalable and efficient means of quantifying weight-space uncertainty in deep neural networks, enabling the application of Bayesian tools such as predictive uncertainty and model selection via Occam's razor. In this work, we introduce laplax, a new open-source Python package for performing Laplace approximations with jax. Designed with a modular and purely functional architecture and minimal external dependencies, laplax offers a flexible and researcher-friendly framework for rapid prototyping and experimentation. Its goal is to facilitate research on Bayesian neural networks, uncertainty quantification for deep learning, and the development of improved Laplace approximation techniques. |
| title | laplax -- Laplace Approximations with JAX |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2507.17013 |