Sparse canonical correlation analysis

By David Roi Hardoon, John Shawe-Taylor. Machine-mediated learning, 2009

325 citations recorded by the source corpus.

We present a novel method for solving Canonical Correlation Analysis (CCA) in a sparse convex framework using a least squares approach. The presented method focuses on the scenario when one is interested in (or limited to) a primal representation for the first view while having a dual representation for the second view. Sparse CCA (SCCA) minimises the number of features used in both the primal and dual projections while maximising the correlation between the two views. The method is compared to alternative sparse solutions as well as demonstrated on paired corpuses for mate-retrieval. We are able to observe, in the mate-retrieval, that when the number of the original features is large SCCA outperforms Kernel CCA (KCCA), learning the common semantic space from a sparse set of features.

Subfields: Computer Vision and Pattern Recognition, Media Technology, Signal Processing

View Paper: https://www.semanticscholar.org/paper/80928f589c1ae6bed0348026f081fc7248a5df66

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

In Signal Processing

Citations in the corpus

Integration strategies of multi-omics data for machine learning analysis (2021, methodology): “Sparse FE methods can be used for feature selection and include Sparse PCA (sPCA) [58], Sparse Canonical Correlation Analysis (CCA) [59], Sparse Non-Negative Matrix Factorization (Sparse NMF) [60], Sparse CA [61], etc.”
Multi-view learning overview: Recent progress and new challenges (2017, methodology): “CCA has been extended to sparse CCA [74, 75] and has been widely used for multi-view classification [76], clustering [77], regression [78], etc.”
A survey of multi-view machine learning (2013, background): “It was further extended to nonlinear subspace learning [3, 17] and sparse formulations [2, 10, 20].”
Sparse probabilistic projections (2008, background): “Recent attempts for constructing sparse CCA include [10, 19].”
Integrated Omics: Tools, Advances, and Future Approaches. (2019, methodology): “Other omics integrative frameworks involve sparse canonical correlation 456 analysis (Parkhomenko et al. 2009), multiple factor analysis (De Tayrac et al.”
A Survey of Multi-View Representation Learning (2016, unknown): “By formulating CCA as two constrained simultaneous regression problems, Hardoon and ShaweTaylor [73] propose to approximate the non-convex constraints with ∞-norm.”
Canonical Correlation Analysis based on Hilbert-Schmidt Independence Criterion and Centered Kernel Target Alignment (2013, background): “More recently, the research community proposed regularized approaches to CCA for simultaneous analysis of multiple high-dimension data sets (Witten & Tibshirani, 2009; Parkhomenko et al., 2009; Hardoon & Shawe-Taylor, 2011).”
Random matrix theory in statistics: A review (2014, methodology): “…for the X and Y variables in the canonical correlation computation), are constrained to be sparse through imposition of sparse penalty, have also been proposed in the literature, notably by Hardoon and Shawe-Taylor (2009), Lee et al. (2009), Parkhomenko et al. (2009) and Witten et al. (2009).”