Generalized Constrained Redundancy Analysis

10 citations recorded by the source corpus.

How researchers see this paper

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

Accelerating Very Deep Convolutional Networks for Classification and Detection (2015, background): “Although the former (“decomposition”) attracts more attention because it directly addresses the time complexity, the latter (“optimization”) is also essential because not all decompositions are equally easy to fine good local optima.”
Efficient and accurate approximations of nonlinear convolutional networks (2014, methodology): “The GSVD is applied on M̂ as M̂ = USV>, such that U is a d-by-d orthogonal matrix satisfying U>U = Id where Id is a d-by-d identity matrix, and V is a d-by-d matrix satisfying V>YY>V = Id (called generalized orthogonality).”
Taxonomy and Evaluation of Structured Compression of Convolutional Neural Networks (2019, unknown): “After combining responses (yi − y), and (zi − z) into matrices Ŷ and Ẑ, the minimization can be written as: argmin M ∥∥∥Ŷ −MẐ∥∥∥2 F s.t. rank M ≤ r. (27) The problem has a closed form solution for M based on generalized SVD (Takane and Jung, 2006).”
Cascaded Projection: End-To-End Network Compression and Acceleration (2019, methodology): “We obtain the transformation Φ(·) that projects each proxy matrix Xi to the closest location on the Grassmann manifold by performing Singular Value Decomposition (SVD) on Xi, such that Xi=UiΣiVTi , where Ui and VTi are orthogonal matrices and Σi is the matrix of singular values.”