Convex Learning with Invariances

Abstract

Incorporating invariances into a learning algorithm is a common problem in ma- chine learning. We provide a convex formulation which can deal with arbitrary loss functions and arbitrary losses. In addition, it is a drop-in replacement for most optimization algorithms for kernels, including solvers of the SVMStruct family. The advantage of our setting is that it relies on column generation instead of mod- ifying the underlying optimization problem directly.

Cite

Text

Teo et al. "Convex Learning with Invariances." Neural Information Processing Systems, 2007.

Markdown

[Teo et al. "Convex Learning with Invariances." Neural Information Processing Systems, 2007.](https://mlanthology.org/neurips/2007/teo2007neurips-convex/)

BibTeX

@inproceedings{teo2007neurips-convex,
  title     = {{Convex Learning with Invariances}},
  author    = {Teo, Choon H. and Globerson, Amir and Roweis, Sam T. and Smola, Alex J.},
  booktitle = {Neural Information Processing Systems},
  year      = {2007},
  pages     = {1489-1496},
  url       = {https://mlanthology.org/neurips/2007/teo2007neurips-convex/}
}