Empirische Inferenz Conference Paper 2004

Multivariate Regression via Stiefel Manifold Constraints

We introduce a learning technique for regression between high-dimensional spaces. Standard methods typically reduce this task to many one-dimensional problems, with each output dimension considered independently. 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, and dependence between the outputs can be exploited.

Author(s): BakIr, G. and Gretton, A. and Franz, M. and Schölkopf, B.
Book Title: Lecture Notes in Computer Science, Vol. 3175
Journal: Pattern Recognition, Proceedings of the 26th DAGM Symposium
Pages: 262-269
Year: 2004
Day: 0
Editors: CE Rasmussen and HH B{\"u}lthoff and B Sch{\"o}lkopf and MA Giese
Publisher: Springer
Bibtex Type: Conference Paper (inproceedings)
Address: Berlin, Germany
Event Name: Pattern Recognition, Proceedings of the 26th DAGM Symposium
Event Place: Tübingen, Germany
Digital: 0
Electronic Archiving: grant_archive
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik
Links:

BibTex

@inproceedings{2845,
  title = {Multivariate Regression via Stiefel Manifold Constraints},
  journal = {Pattern Recognition, Proceedings of the 26th DAGM Symposium},
  booktitle = {Lecture Notes in Computer Science, Vol. 3175},
  abstract = {We introduce a learning technique for regression
  between high-dimensional spaces. Standard methods typically reduce
  this task to many one-dimensional problems, with each output
  dimension considered independently. 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, and dependence between the
  outputs can be exploited.},
  pages = {262-269},
  editors = {CE Rasmussen and  HH B{\"u}lthoff and B Sch{\"o}lkopf and MA Giese},
  publisher = {Springer},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  address = {Berlin, Germany},
  year = {2004},
  slug = {2845},
  author = {BakIr, G. and Gretton, A. and Franz, M. and Sch{\"o}lkopf, B.}
}