论文信息 - Boosting for Control of Dynamical Systems

Boosting for Control of Dynamical Systems

We study the question of how to aggregate controllers for dynamical systems in order to improve their performance. To this end, we propose a framework of boosting for online control. Our main result is an efficient boosting algorithm that combines weak controllers into a provably more accurate one. Empirical evaluation on a host of control settings supports our theoretical findings.

Naman Agarwal | Elad Hazan | Zhou Lu | Nataly Brukhim

[1] Yoram Singer,et al. An Efficient Boosting Algorithm for Combining Preferences by , 2013 .

[2] Yoav Freund,et al. A decision-theoretic generalization of on-line learning and an application to boosting , 1995, EuroCOLT.

[3] Shie Mannor,et al. Online Learning for Adversaries with Memory: Price of Past Mistakes , 2015, NIPS.

[4] Alex M. Andrew,et al. Boosting: Foundations and Algorithms , 2012 .

[5] Sham M. Kakade,et al. Online Control with Adversarial Disturbances , 2019, ICML.

[6] Sham M. Kakade,et al. The Nonstochastic Control Problem , 2020, ALT.

[7] Wojciech Zaremba,et al. OpenAI Gym , 2016, ArXiv.

[8] Yishay Mansour,et al. Learning Linear-Quadratic Regulators Efficiently with only $\sqrt{T}$ Regret , 2019, ICML.

[9] Haipeng Luo,et al. Optimal and Adaptive Algorithms for Online Boosting , 2015, ICML.

[10] Haipeng Luo,et al. Online Gradient Boosting , 2015, NIPS.

[11] Max Simchowitz,et al. Improper Learning for Non-Stochastic Control , 2020, COLT.

[12] Adam Tauman Kalai,et al. Potential-Based Agnostic Boosting , 2009, NIPS.

[13] Avinatan Hassidim,et al. Online Linear Quadratic Control , 2018, ICML.

[14] Peter L. Bartlett,et al. Boosting Algorithms as Gradient Descent , 1999, NIPS.