2020/11/03 by Elena Tuzhilina, Tuzhilina, Elena, Leonardo Tozzi +3 · 2 citations
Biochemistry, Genetics and Molecular Biology · Mathematics · #FOS: Computer and information sciences #Fractal and DNA sequence analysis #Gene expression and cancer classification #Methodology (stat.ME) #Morphological variations and asymmetry
paper · pdf · doi:10.48550/arxiv.2011.01650
openalex publication_date 2020/11/03 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Canonical correlation analysis (CCA) is a technique for measuring the\nassociation between two multivariate data matrices. A regularized modification\nof canonical correlation analysis (RCCA) which imposes an \ℓ2 penalty on\nthe CCA coefficients is widely used in applications with high-dimensional data.\nOne limitation of such regularization is that it ignores any data structure,\ntreating all the features equally, which can be ill-suited for some\napplications. In this paper we introduce several approaches to regularizing CCA\nthat take the underlying data structure into account. In particular, the\nproposed group regularized canonical correlation analysis (GRCCA) is useful\nwhen the variables are correlated in groups. We illustrate some computational\nstrategies to avoid excessive computations with regularized CCA in high\ndimensions. We demonstrate the application of these methods in our motivating\napplication from neuroscience, as well as in a small simulation example.\n