Bayesian mixture of gaussian processes for data association problem

作者：

Highlights：

• A probabilistic model using mixture of Gaussian processes with a Bayesian approaches for a data association problem.

• The number of hyper-parameters is decreased by using a new EM algorithm, leading to choose better hyper-parameters.

• The algorithm gives both an association of observations and an estimation of global mixture weights of latent sources.

• We derive a lower bound based on theoretical analysis to estimate the effect of the proposed algorithm.

摘要

•A probabilistic model using mixture of Gaussian processes with a Bayesian approaches for a data association problem.•The number of hyper-parameters is decreased by using a new EM algorithm, leading to choose better hyper-parameters.•The algorithm gives both an association of observations and an estimation of global mixture weights of latent sources.•We derive a lower bound based on theoretical analysis to estimate the effect of the proposed algorithm.

论文关键词：Gaussian processes,Bayesian models,Variational inference,Expectation maximization

论文评审过程：Received 17 April 2021, Revised 13 January 2022, Accepted 16 February 2022, Available online 19 February 2022, Version of Record 26 February 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108592