arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 1605.03364math.NAcs.LGcs.NAstat.ML
收藏

在贝叶斯微分方程求解器中的主动不确定性校准

Active Uncertainty Calibration in Bayesian ODE Solvers

Max-Planck-Institute for Intelligent Systems(马克斯·普朗克智能系统研究所)

Hans Kersting, Philipp Hennig

AI总结:

本文研究了如何在贝叶斯微分方程求解器中平衡计算成本与概率校准,提出了一种基于过滤的方法Bayesian Quadrature filtering (BQF),通过主动学习梯度测量的不精确性来提高不确定性校准。

摘要:

在统计学和机器学习中,对微分方程(ODEs)求解器的兴趣正在重新增长,这些求解器返回概率测度而非点估计。最近,Conrad等人引入了一种基于采样的方法类,这些方法在特定意义上是'well-calibrated'的。但是,这些方法的计算成本显著高于经典方法。另一方面,Schober等人指出经典Runge-Kutta ODE求解器与高斯滤波器之间存在精确的联系,这只能提供粗糙的概率校准,但计算开销可忽略不计。通过将ODE的解视为线性高斯SDE中的近似推断,我们研究了一类概率ODE求解器,这些求解器在计算成本和概率校准之间取得了平衡,并识别出不准确的梯度测量是不确定性的关键来源。我们提出了一种新的基于过滤的方法Bayesian Quadrature filtering (BQF),该方法利用贝叶斯二次法主动学习梯度测量的不精确性,通过收集多个梯度评估来提高不确定性校准。

英文摘要:

There is resurging interest, in statistics and machine learning, in solvers for ordinary differential equations (ODEs) that return probability measures instead of point estimates. Recently, Conrad et al. introduced a sampling-based class of methods that are 'well-calibrated' in a specific sense. But the computational cost of these methods is significantly above that of classic methods. On the other hand, Schober et al. pointed out a precise connection between classic Runge-Kutta ODE solvers and Gaussian filters, which gives only a rough probabilistic calibration, but at negligible cost overhead. By formulating the solution of ODEs as approximate inference in linear Gaussian SDEs, we investigate a range of probabilistic ODE solvers, that bridge the trade-off between computational cost and probabilistic calibration, and identify the inaccurate gradient measurement as the crucial source of uncertainty. We propose the novel filtering-based method Bayesian Quadrature filtering (BQF) which uses Bayesian quadrature to actively learn the imprecision in the gradient measurement by collecting multiple gradient evaluations.

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。

↑