Bayesian Change Point Detection with Spike-and-Slab Priors
Bayesian Change Point Detection with Spike-and-Slab Priors
复制标题
DOI：
10.1080/10618600.2023.2182312
复制
发表时间：
2023-02
影响因子：
2.4
通讯作者：
L. Cappello;Oscar Hernan Madrid Padilla;Julia A. Palacios
中科院分区：
文献类型：
--
作者：
L. Cappello;Oscar Hernan Madrid Padilla;Julia A. Palacios
摘要：本文研究了用尖峰-板先验方法对变化点的数目及其位置进行一致估计。利用变量选择文献中的最新结果，我们证明了基于spike- slab先验的估计器在多个离线变化点检测问题中实现了最佳定位率。在此基础上，提出了一种最快的贝叶斯方法——贝叶斯变化点检测方法。我们通过实证工作证明了我们的方法相对于一些最先进的基准的良好性能。有趣的是，尽管有高斯噪声假设，我们的方法在数值实验中对误差项的错误说明比竞争方法更具鲁棒性。本文的补充材料可在网上获得。
Abstract We study the use of spike-and-slab priors for consistent estimation of the number of change points and their locations. Leveraging recent results in the variable selection literature, we show that an estimator based on spike-and-slab priors achieves optimal localization rate in the multiple offline change point detection problem. Based on this estimator, we propose a Bayesian change point detection method, which is one of the fastest Bayesian methodologies. We demonstrate through empirical work the good performance of our approach vis-a-vis some state-of-the-art benchmarks. Interestingly, despite having a Gaussian noise assumption, our approach is more robust to misspecification of the error terms than the competing methods in numerical experiments. Supplementary materials for this article are available online.