Parallel Affine Transformation Tuning of Markov Chain Monte Carlo
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929447771308032 |
|---|---|
| author | Schär, Philip Habeck, Michael Rudolf, Daniel |
| author_facet | Schär, Philip Habeck, Michael Rudolf, Daniel |
| contents | The performance of Markov chain Monte Carlo samplers strongly depends on the properties of the target distribution such as its covariance structure, the location of its probability mass and its tail behavior. We explore the use of bijective affine transformations of the sample space to improve the properties of the target distribution and thereby the performance of samplers running in the transformed space. In particular, we propose a flexible and user-friendly scheme for adaptively learning the affine transformation during sampling. Moreover, the combination of our scheme with Gibbsian polar slice sampling is shown to produce samples of high quality at comparatively low computational cost in several settings based on real-world data. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_16567 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Parallel Affine Transformation Tuning of Markov Chain Monte Carlo Schär, Philip Habeck, Michael Rudolf, Daniel Methodology Machine Learning 60J22, 65C05, 60J05 The performance of Markov chain Monte Carlo samplers strongly depends on the properties of the target distribution such as its covariance structure, the location of its probability mass and its tail behavior. We explore the use of bijective affine transformations of the sample space to improve the properties of the target distribution and thereby the performance of samplers running in the transformed space. In particular, we propose a flexible and user-friendly scheme for adaptively learning the affine transformation during sampling. Moreover, the combination of our scheme with Gibbsian polar slice sampling is shown to produce samples of high quality at comparatively low computational cost in several settings based on real-world data. |
| title | Parallel Affine Transformation Tuning of Markov Chain Monte Carlo |
| topic | Methodology Machine Learning 60J22, 65C05, 60J05 |
| url | https://arxiv.org/abs/2401.16567 |