Multivariate bounded support asymmetric generalized Gaussian mixture model with model selection using minimum message length
作者：
Highlights：
• Bounded support asymmetric generalized Gaussian mixture model (BAGGMM) is proposed.
• Parameters estimation is performed through ML and EM with Newton Raphson algorithm.
• Model is validated via image spam detection, object & visual scene categorization.
• Model selection criterion for BAGGMM using Minimum Message Length (MML) is proposed.
• Effectiveness of MML in proposed BAGGMM is tested via above mentioned applications.
摘要
•Bounded support asymmetric generalized Gaussian mixture model (BAGGMM) is proposed.•Parameters estimation is performed through ML and EM with Newton Raphson algorithm.•Model is validated via image spam detection, object & visual scene categorization.•Model selection criterion for BAGGMM using Minimum Message Length (MML) is proposed.•Effectiveness of MML in proposed BAGGMM is tested via above mentioned applications.
论文关键词：Multivariate bounded asymmetric generalized Gaussian mixture model (BAGGMM),Minimum message length (MML),Model selection,Data clustering,Expectation–maximization (EM),Newton Raphson
论文评审过程：Received 11 April 2020, Revised 19 April 2022, Accepted 4 May 2022, Available online 18 May 2022, Version of Record 23 May 2022.
论文官网地址：https://doi.org/10.1016/j.eswa.2022.117516
