A PAC-Bayes Approach for Controlling Unknown Linear Discrete-time Systems
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914584355405824 |
|---|---|
| author | Luo, Yujia Pu, Ye Manton, Jonathan H. Zhu, Jingge |
| author_facet | Luo, Yujia Pu, Ye Manton, Jonathan H. Zhu, Jingge |
| contents | This paper presents a PAC-Bayes framework for learning controllers for unknown stochastic linear discrete-time systems, where the system parameters are drawn from a fixed but unknown distribution. We derive a data-dependent high probability bound on the performance of any learned (stochastic) controller, and propose novel efficient learning algorithms with theoretical guarantees, which can be implemented for both finite and infinite controller spaces. Compared to prior work, our bound holds for unbounded quadratic cost. In the special case where LQG is optimal, our numerical results suggest that the learned controllers achieve comparable performance to LQG. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_10493 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | A PAC-Bayes Approach for Controlling Unknown Linear Discrete-time Systems Luo, Yujia Pu, Ye Manton, Jonathan H. Zhu, Jingge Optimization and Control Systems and Control Machine Learning This paper presents a PAC-Bayes framework for learning controllers for unknown stochastic linear discrete-time systems, where the system parameters are drawn from a fixed but unknown distribution. We derive a data-dependent high probability bound on the performance of any learned (stochastic) controller, and propose novel efficient learning algorithms with theoretical guarantees, which can be implemented for both finite and infinite controller spaces. Compared to prior work, our bound holds for unbounded quadratic cost. In the special case where LQG is optimal, our numerical results suggest that the learned controllers achieve comparable performance to LQG. |
| title | A PAC-Bayes Approach for Controlling Unknown Linear Discrete-time Systems |
| topic | Optimization and Control Systems and Control Machine Learning |
| url | https://arxiv.org/abs/2605.10493 |