Multi-view Regression Via Canonical Correlation Analysis

By Sham M. Kakade, Dean Phillips Foster. 2007

221 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Statistics and Probability, Representation & Self-Supervised Learning

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 Statistics and Probability

In Representation & Self-Supervised Learning

Citations in the corpus

A Survey on Multi-view Learning (2013, methodology): “They have been further developed to conduct multi-view clustering (Chaudhuri et al., 2009) and regression (Kakade and Foster, 2007).”
Multi-Stream Deep Neural Networks for RGB-D Egocentric Action Recognition (2019, background): “This is because DCCA only utilizes the sharable information of different modalities, while the MDNN further preserves their distinctive characteristics.”
Unsupervised Risk Estimation with only Structural Assumptions (2016, methodology): “Our work does not strictly generalize the work above, as for instance Kakade & Foster (2007) and Blitzer et al. (2011) assume low regret but not independence, and consider regression rather than classification.”
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 Representation Learning (2016, unknown): “regression [48, 49], word embeddings [50–52], and discriminant learning [53–55].”
Multi-View Learning in the Presence of View Disagreement (2008, background): “Multi-view learning can be advantageous when compared to learning with only a single view [3, 4, 12], especially when the weaknesses of one view complement the strengths of the other.”
From Dependence to Causation (2016, background): “CCA has likewise found numerous applications, including multi-view statistics (Kakade and Foster, 2007) and learning with missing features (Chaudhuri et al., 2009; Lopez-Paz et al., 2014).”
Deep Canonical Correlation Analysis (2013, background): “…learning features for a single view when another view is available for representation learning but not at prediction time (Blaschko & Lampert, 2008; Chaudhuri et al., 2009; Arora & Livescu, 2012); and reducing sample complexity of prediction problems using unlabeled data (Kakade & Foster, 2007).”
Multiview Consensus Graph Clustering (2019, methodology): “Besides using co-training strategy and HSIC, canonical correlation analysis is used to exploit feature correlation between different views [20], [21].”
A Spectral Algorithm for Latent Dirichlet Allocation (2012, background): “Some preliminary experiments that illustrate the efficacy of the proposed algorithm are given in the appendix.”
Randomized Nonlinear Component Analysis (2014, methodology): “CCA is widely used to learn from multiple modalities of data (Kakade & Foster, 2007), an ability particularly useful when some of the modalities are only available at training time but keeping information about them at testing time is beneficial (Chaudhuri et al., 2009; Vapnik & Vashist, 2009).”
Posterior Regularization for Structured Latent Variable Models (2010, background): “This regularization can lead to better convergence by signiﬁcantly decreasing the effective size of our hypothesis class (Balcan and Blum, 2005; Kakade and Foster, 2007; Rosenberg and Bartlett, 2007).”