arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1802.00852stat.MEcs.NAmath.NA
收藏

利用代理随机过程对动态系统进行参数和不确定性估计

Parameter and Uncertainty Estimation for Dynamical Systems Using Surrogate Stochastic Processes

Academy of Integrated Science, Virginia Tech(弗吉尼亚理工大学综合科学学院);Booth School of Business, The University of Chicago(芝加哥大学布斯商学院);Department of Statistics, Virginia Tech(弗吉尼亚理工大学统计系);Department of Internal Medicine, University of Tennessee Health Science Center(田纳西健康科学中心内科系);Department of Pediatrics, University of Tennessee Health Science Center(田纳西健康科学中心儿科系)

M. Chung, M. Binois, R. B. Gramacy, D. J. Moquin, A. P. Smith, A. M. Smith

AI总结:

本文提出了一种新的方法,通过观测数据学习参数化的动态系统。首先拟合代理随机过程以融入先验知识,然后利用其样本作为代理数据进行点估计,该方法具有全贝叶斯性和并行计算能力。

摘要:

通过观测数据对动态系统中未知量进行推断对于提供有意义的见解、提供准确预测、实现稳健控制和建立未来实验设计至关重要。将数学理论与实证测量在统计上一致的方式结合是关键,但存在诸多挑战,例如参数估计问题的不恰当性、适当正则化和整合先验知识,以及对完整不确定性资格的计算限制。为了解决这些问题,我们提出了一种新的方法,从数据中学习参数化的动态系统。在许多方面,我们的方法颠倒了传统框架。我们首先拟合一个代理随机过程以适应观测数据,强制先验知识(例如光滑性),并处理具有挑战性的数据特征,如异方差性、厚尾和删失。然后,随机过程的样本被用作“代理数据”,通过普通点估计方法以模块化的方式计算点估计。这种方法的一个优点是完全贝叶斯且同时可并行化。我们在捕食者-猎物模拟研究和一个涉及宿主内流感病毒感染数据与病毒动力学模型的现实应用中展示了新方法的优势。

英文摘要:

Inference on unknown quantities in dynamical systems via observational data is essential for providing meaningful insight, furnishing accurate predictions, enabling robust control, and establishing appropriate designs for future experiments. Merging mathematical theory with empirical measurements in a statistically coherent way is critical and challenges abound, e.g.,: ill-posedness of the parameter estimation problem, proper regularization and incorporation of prior knowledge, and computational limitations on full uncertainty qualification. To address these issues, we propose a new method for learning parameterized dynamical systems from data. In many ways, our proposal turns the canonical framework on its head. We first fit a surrogate stochastic process to observational data, enforcing prior knowledge (e.g., smoothness), and coping with challenging data features like heteroskedasticity, heavy tails and censoring. Then, samples of the stochastic process are used as "surrogate data" and point estimates are computed via ordinary point estimation methods in a modular fashion. An attractive feature of this approach is that it is fully Bayesian and simultaneously parallelizable. We demonstrate the advantages of our new approach on a predator prey simulation study and on a real world application involving within-host influenza virus infection data paired with a viral kinetic model.

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。

↑