Large Margin vs. Large Volume in Transductive Learning

Abstract

We consider a large volume principle for transductive learning that prioritizes the transductive equivalence classes according to the volume they occupy in hypothesis space. We approximate volume maximization using a geometric interpretation of the hypothesis space. The resulting algorithm is defined via a non-convex optimization problem that can still be solved exactly and efficiently. We provide a bound on the test error of the algorithm and compare it to transductive SVM (TSVM) using 31 datasets.

Cite

Text

El-Yaniv et al. "Large Margin vs. Large Volume in Transductive Learning." Machine Learning, 2008. doi:10.1007/S10994-008-5071-9

Markdown

[El-Yaniv et al. "Large Margin vs. Large Volume in Transductive Learning." Machine Learning, 2008.](https://mlanthology.org/mlj/2008/elyaniv2008mlj-large/) doi:10.1007/S10994-008-5071-9

BibTeX

@article{elyaniv2008mlj-large,
  title     = {{Large Margin vs. Large Volume in Transductive Learning}},
  author    = {El-Yaniv, Ran and Pechyony, Dmitry and Vapnik, Vladimir},
  journal   = {Machine Learning},
  year      = {2008},
  pages     = {173-188},
  doi       = {10.1007/S10994-008-5071-9},
  volume    = {72},
  url       = {https://mlanthology.org/mlj/2008/elyaniv2008mlj-large/}
}