Mixture of grouped regressors and its application to visual mapping
作者：
Highlights：
• Mixture of grouped regressors (MoGR) is proposed to tackle overfitting problems of existing Mixture of Regressors methods.
• MoGR partitions the individual regressors in mixture regression model into a number of groups.
• The parameters of each regressor are learned using all data within a group, rather than a cluster.
• MoGR requires a small number of training data and is robust to noise.
• It has shown obviously improved performance when compared with state-of-the-art nonlinear visual mapping methods.
摘要
Highlights•Mixture of grouped regressors (MoGR) is proposed to tackle overfitting problems of existing Mixture of Regressors methods.•MoGR partitions the individual regressors in mixture regression model into a number of groups.•The parameters of each regressor are learned using all data within a group, rather than a cluster.•MoGR requires a small number of training data and is robust to noise.•It has shown obviously improved performance when compared with state-of-the-art nonlinear visual mapping methods.
论文关键词：Mixture of regressors,Group partition,EM algorithm
论文评审过程：Received 25 April 2014, Revised 25 September 2015, Accepted 19 October 2015, Available online 11 November 2015, Version of Record 8 February 2016.
论文官网地址：https://doi.org/10.1016/j.patcog.2015.10.016
