2016/01/14 by Hien D. Nguyen, Geoffrey J. McLachlan, Nguyen, Hien D +5
Chemistry · Computer Science · Engineering · #Bayesian Methods and Mixture Models #FOS: Computer and information sciences #Methodology (stat.ME) #Remote-Sensing Image Classification #Spectroscopy and Chemometric Analyses #Time Series Analysis and Forecasting
paper · pdf · doi:10.48550/arxiv.1601.03517
openalex publication_date 2016/01/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Time-series data arise in many medical and biological imaging scenarios. In\nsuch images, a time-series is obtained at each of a large number of\nspatially-dependent data units. It is interesting to organize these data into\nmodel-based clusters. A two-stage procedure is proposed. In Stage 1, a mixture\nof autoregressions (MoAR) model is used to marginally cluster the data. The\nMoAR model is fitted using maximum marginal likelihood (MMaL) estimation via an\nMM (minorization--maximization) algorithm. In Stage 2, a Markov random field\n(MRF) model induces a spatial structure onto the Stage 1 clustering. The MRF\nmodel is fitted using maximum pseudolikelihood (MPL) estimation via an MM\nalgorithm. Both the MMaL and MPL estimators are proved to be consistent.\nNumerical properties are established for both MM algorithms. A simulation study\ndemonstrates the performance of the two-stage procedure. An application to the\nsegmentation of a zebrafish brain calcium image is presented.\n