Sparse Generalised Principal Component Analysis. (November 2018)

• Record Type: Journal Article
• Authors: Smallman, Luke; Artemiou, Andreas; Morgan, Jennifer
• Is Part Of: Pattern recognition. Volume 83(2018:Nov.)
• Page Start: 443
• Page End: 455
• Publication Date: 2018-11
• Subjects: Dimension reduction -- PCA -- Text mining -- Exponential family; 62H25 -- 62-09 -- 62J07 -- 68T50; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2018.06.014
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A method for sparse feature extraction of exponential family data is presented. It extends the previous method of Generalised Principal Component Analysis. Applications to text data are performed using a simple Poisson model. Superior performance to current state of art methodology is shown on synthetic data. Performance on a dataset from healthcare is on par with state of art methodology. Abstract: In this paper, we develop a sparse method for unsupervised dimension reduction for data from an exponential-family distribution. Our idea extends previous work on Generalised Principal Component Analysis by adding L 1 and SCAD penalties to introduce sparsity. We demonstrate the significance and advantages of our method with synthetic and real data examples. We focus on the application to text data which is high-dimensional and non-Gaussian by nature and discuss the potential advantages of our methodology in achieving dimension reduction.