A Particle Algorithm for Mean-Field Variational Inference
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917175952932864 |
|---|---|
| author | Du, Qiang Wang, Kaizheng Zhang, Edith Zhong, Chenyang |
| author_facet | Du, Qiang Wang, Kaizheng Zhang, Edith Zhong, Chenyang |
| contents | Variational inference is a fast and scalable alternative to Markov chain Monte Carlo and has been widely applied to posterior inference tasks in statistics and machine learning. A traditional approach for implementing mean-field variational inference (MFVI) is coordinate ascent variational inference (CAVI), which relies crucially on parametric assumptions on complete conditionals. We introduce a novel particle-based algorithm for MFVI, named PArticle VI (PAVI), for nonparametric mean-field approximation. We obtain non-asymptotic error bounds for our algorithm. To our knowledge, this is the first end-to-end guarantee for particle-based MFVI. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_20385 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Particle Algorithm for Mean-Field Variational Inference Du, Qiang Wang, Kaizheng Zhang, Edith Zhong, Chenyang Statistics Theory Machine Learning Optimization and Control 65K10, 65C35, 62F15, 60J60, 90C15 Variational inference is a fast and scalable alternative to Markov chain Monte Carlo and has been widely applied to posterior inference tasks in statistics and machine learning. A traditional approach for implementing mean-field variational inference (MFVI) is coordinate ascent variational inference (CAVI), which relies crucially on parametric assumptions on complete conditionals. We introduce a novel particle-based algorithm for MFVI, named PArticle VI (PAVI), for nonparametric mean-field approximation. We obtain non-asymptotic error bounds for our algorithm. To our knowledge, this is the first end-to-end guarantee for particle-based MFVI. |
| title | A Particle Algorithm for Mean-Field Variational Inference |
| topic | Statistics Theory Machine Learning Optimization and Control 65K10, 65C35, 62F15, 60J60, 90C15 |
| url | https://arxiv.org/abs/2412.20385 |