arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1907.05451cs.PL
收藏

具有收敛保证的组合推理元程序设计

Compositional Inference Metaprogramming with Convergence Guarantees

Shivam Handa, Vikash Mansinghka, Martin Rinard

摘要:

推理元编程通过支持将概率程序的执行分解为子问题和将不同的概率推理算法应用于不同子问题的混合概率推理算法的部署来实现有效的概率编程。我们引入了独立子问题推理的概念(相对于纠缠子问题推理,其中的子问题推理算法在整个程序跟踪操作),并提出了一个数学框架,用于研究收敛性能的混合推理算法,适用于不同的马尔可夫链蒙特卡罗算法的推理问题的不同部分。然后,我们使用这种形式主义证明渐近收敛结果的概率程序推理元编程。据我们所知,这是第一个渐近收敛的混合概率推理算法定义(基于子问题)推理元编程的结果。

英文摘要:

Inference metaprogramming enables effective probabilistic programming by supporting the decomposition of executions of probabilistic programs into subproblems and the deployment of hybrid probabilistic inference algorithms that apply different probabilistic inference algorithms to different subproblems. We introduce the concept of independent subproblem inference (as opposed to entangled subproblem inference in which the subproblem inference algorithm operates over the full program trace) and present a mathematical framework for studying convergence properties of hybrid inference algorithms that apply different Markov-Chain Monte Carlo algorithms to different parts of the inference problem. We then use this formalism to prove asymptotic convergence results for probablistic programs with inference metaprogramming. To the best of our knowledge this is the first asymptotic convergence result for hybrid probabilistic inference algorithms defined by (subproblem-based) inference metaprogramming.

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。

↑