Data-free metrics for Dirichlet and generalized Dirichlet mixture-based HMMs – A practical study
作者：
Highlights：
• Proposition of a new similarity measure for Dirichlet and generalized Dirichlet HMMs (two variants).
• Not trivial generalization of existing parametric similarity measure.
• Proposition of quality scores for performance characterization of the similarity measures.
• Extensive experiments on synthetic data highlighting the performance on different aspects of the newly proposed and state-of-the-art measures.
• Illustration of newly proposed similarity measure performance on real-world data sets.
摘要
•Proposition of a new similarity measure for Dirichlet and generalized Dirichlet HMMs (two variants).•Not trivial generalization of existing parametric similarity measure.•Proposition of quality scores for performance characterization of the similarity measures.•Extensive experiments on synthetic data highlighting the performance on different aspects of the newly proposed and state-of-the-art measures.•Illustration of newly proposed similarity measure performance on real-world data sets.
论文关键词：Hidden Markov models,Similarity measure,Dirichlet,Generalized Dirichlet
论文评审过程：Received 17 February 2018, Revised 30 June 2018, Accepted 27 August 2018, Available online 27 August 2018, Version of Record 8 September 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.08.013
