首页> 外文OA文献 >Finite state space non parametric Hidden Markov Models are in general identifiable
【2h】

Finite state space non parametric Hidden Markov Models are in general identifiable

机译:有限状态空间非参数隐马尔可夫模型通常是   识别

摘要

In this paper, we prove that finite state space non parametric hidden Markovmodels are identifiable as soon as the transition matrix of the latent Markovchain has full rank and the emission probability distributions are linearlyindependent. We then propose several non parametric likelihood based estimationmethods, which we apply to models used in applications. We finally show onexamples that the use of non parametric modeling and estimation may improve theclassification performances.
机译:在本文中,我们证明了只要隐马尔可夫链的转移矩阵满秩且发射概率分布与线性无关,就可以识别出有限状态空间非参数隐马尔可夫模型。然后,我们提出了几种基于非参数似然的估计方法,这些方法适用于应用程序中使用的模型。我们最终在示例中显示了使用非参数建模和估计可以改善分类性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号