arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1906.03028stat.MLcs.LGcs.PL
收藏

概率程序的自动重新参数化

Automatic Reparameterisation of Probabilistic Programs

University of Edinburgh(爱丁堡大学);Google(谷歌)

Maria I. Gorinova, Dave Moore, Matthew D. Hoffman

摘要:

概率编程已经成为统计学、应用科学和机器学习中一个强大的范例:通过将建模与推理分离,它承诺允许建模者直接推理生成数据的过程。然而,推理算法的性能可能会受到用于表达模型的参数化的显着影响,需要用户以非直观的方式转换他们的程序。我们主张自动化这些转换,并证明,在最近的建模框架中可用的机制可以实现非中心化和相关的reparameterisations。这使得新的推理算法,我们提出了两个:一个简单的方法,使用交错采样和一个新的变分公式,搜索在一个连续的参数化空间。我们表明,这些方法可以在一系列模型中进行稳健的推断,并且可以产生比最佳固定参数化更有效的采样器。

英文摘要:

Probabilistic programming has emerged as a powerful paradigm in statistics, applied science, and machine learning: by decoupling modelling from inference, it promises to allow modellers to directly reason about the processes generating data. However, the performance of inference algorithms can be dramatically affected by the parameterisation used to express a model, requiring users to transform their programs in non-intuitive ways. We argue for automating these transformations, and demonstrate that mechanisms available in recent modeling frameworks can implement non-centring and related reparameterisations. This enables new inference algorithms, and we propose two: a simple approach using interleaved sampling and a novel variational formulation that searches over a continuous space of parameterisations. We show that these approaches enable robust inference across a range of models, and can yield more efficient samplers than the best fixed parameterisation.

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。

↑