Learning for non-stationary Dirichlet processes
Learning for non-stationary Dirichlet processes
复制标题
DOI：
10.1002/acs.949
复制
发表时间：
2007-12-01
影响因子：
3.1
通讯作者：
Karny, M.
中科院分区：
文献类型：
--
作者：
Quinn, A.;Karny, M.
Dirichlet过程先验（DPP）用于建模未知的概率分布，F。这消除了对参数模型假设的需要，在存在显著模型不确定性的问题中提供鲁棒性。两个重要的参数技术的学习扩展到这个非参数的情况下，第一次。这些是（i）顺序停止，它提出了使用i.i.d.在线学习F的最佳停止时间。抽样;（ii）稳定的遗忘，更新DPP响应F的变化，但不需要一个正式的过渡模型。在每种情况下，一个实用的和高度听话的算法被发现，并报告了模拟研究。版权所有（C）2007约翰威利父子有限公司
The Dirichlet process prior (DPP) is used to model an unknown probability distribution, F. This eliminates the need for parametric model assumptions, providing robustness in problems where there is significant model uncertainty. Two important parametric techniques for learning are extended to this non-parametric context for the first time. These are (i) sequential stopping, which proposes an optimal stopping time for online learning of F using i.i.d. sampling; and (ii) stabilized forgetting, which updates the DPP in response to changes in F, but without the need for a formal transition model. In each case, a practical and highly tractable algorithm is revealed, and simulation studies are reported. Copyright (C) 2007 John Wiley & Sons, Ltd.