Model-based clustering of high-dimensional data streams with online mixture of probabilistic PCA

摘要: Model-based clustering is a popular tool which renowned for its probabilistic foundations and flexibility. However, model-based techniques usually perform poorly when dealing with high-dimensional data streams, are nowadays frequent type. To overcome this limitation of clustering, we propose an online inference algorithm the mixture PCA model. The proposed relies on EM-based procedure incremental version PCA. Model selection also considered in setting through parallel computing. Numerical experiments simulated real demonstrate effectiveness our approach compare it to state-of-the-art algorithms.