arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1009.3958cs.LGstat.ML
收藏

近似推理与随机最优控制

Approximate Inference and Stochastic Optimal Control

University of Edinburgh(爱丁堡大学);TU Berlin(柏林工业大学)

Konrad Rawlik, Marc Toussaint, Sethu Vijayakumar

摘要:

我们提出了一种新的随机最优控制问题的近似推理问题,证明,这样的解释导致新的实际方法为原来的问题。特别地,我们提出了一类新的迭代解的随机最优控制问题的基础上自然松弛的确切的对偶制定。这些理论见解被应用于强化学习问题,在那里它们导致新的模型自由,关闭离散和连续问题的政策方法。

英文摘要:

We propose a novel reformulation of the stochastic optimal control problem as an approximate inference problem, demonstrating, that such a interpretation leads to new practical methods for the original problem. In particular we characterise a novel class of iterative solutions to the stochastic optimal control problem based on a natural relaxation of the exact dual formulation. These theoretical insights are applied to the Reinforcement Learning problem where they lead to new model free, off policy methods for discrete and continuous problems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑