2015/11/12 by Kai Fan, Katherine Heller, Fan, Kai +1
Computer Science · #Anomaly Detection Techniques and Applications #Bayesian Methods and Mixture Models #FOS: Computer and information sciences #Machine Learning (stat.ML) #Target Tracking and Data Fusion in Sensor Networks #Video Surveillance and Tracking Methods
paper · pdf · doi:10.48550/arxiv.1511.04157
openalex publication_date 2015/11/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
For regular particle filter algorithm or Sequential Monte Carlo (SMC)\nmethods, the initial weights are traditionally dependent on the proposed\ndistribution, the posterior distribution at the current timestamp in the\nsampled sequence, and the target is the posterior distribution of the previous\ntimestamp. This is technically correct, but leads to algorithms which usually\nhave practical issues with degeneracy, where all particles eventually collapse\nonto a single particle. In this paper, we propose and evaluate using k means\nclustering to attack and even take advantage of this degeneracy. Specifically,\nwe propose a Stochastic SMC algorithm which initializes the set of k means,\nproviding the initial centers chosen from the collapsed particles. To fight\nagainst degeneracy, we adjust the regular SMC weights, mediated by cluster\nproportions, and then correct them to retain the same expectation as before. We\nexperimentally demonstrate that our approach has better performance than\nvanilla algorithms.\n