eduzhai > Physical Sciences > Physics Sciences >

Sparse Additive Gaussian Process with Soft Interactions

  • Save

... pages left unread,continue reading

Document pages: 23 pages

Abstract: This paper presents a novel variable selectionmethod in additive nonparametric regression model. This work is motivated bythe need to select the number of nonparametric components and number ofvariables within each nonparametric component. The proposed method uses acombination of hard and soft shrinkages to separately control the number ofadditive components and the variables within each component. An efficientalgorithm is developed to select the importance of variables and estimate theinteraction network. Excellent performance is obtained in simulated and realdata examples.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...