Header logo is

Multivariate Regression with Stiefel Constraints

2004

Technical Report

ei


We introduce a new framework for regression between multi-dimensional spaces. Standard methods for solving this problem typically reduce the problem to one-dimensional regression by choosing features in the input and/or output spaces. These methods, which include PLS (partial least squares), KDE (kernel dependency estimation), and PCR (principal component regression), select features based on different a-priori judgments as to their relevance. Moreover, loss function and constraints are chosen not primarily on statistical grounds, but to simplify the resulting optimisation. By contrast, in our approach the feature construction and the regression estimation are performed jointly, directly minimizing a loss function that we specify, subject to a rank constraint. A major advantage of this approach is that the loss is no longer chosen according to the algorithmic requirements, but can be tailored to the characteristics of the task at hand; the features will then be optimal with respect to this objective. Our approach also allows for the possibility of using a regularizer in the optimization. Finally, by processing the observations sequentially, our algorithm is able to work on large scale problems.

Author(s): Bakir, GH. and Gretton, A. and Franz, MO. and Schölkopf, B.
Number (issue): 128
Year: 2004
Day: 0

Department(s): Empirical Inference
Bibtex Type: Technical Report (techreport)

Institution: MPI for Biological Cybernetics, Spemannstr 38, 72076, Tuebingen

Digital: 0
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik

Links: PDF

BibTex

@techreport{2831,
  title = {Multivariate Regression with Stiefel Constraints},
  author = {Bakir, GH. and Gretton, A. and Franz, MO. and Sch{\"o}lkopf, B.},
  number = {128},
  organization = {Max-Planck-Gesellschaft},
  institution = {MPI for Biological Cybernetics, Spemannstr 38, 72076, Tuebingen},
  school = {Biologische Kybernetik},
  year = {2004},
  doi = {}
}