Multilabel Structured Output Learning with Random Spanning Trees of Max-Margin Markov Networks
Abstract
We show that the usual score function for conditional Markov networks can be written as the expectation over the scores of their spanning trees. We also show that a small random sample of these output trees can attain a significant fraction of the margin obtained by the complete graph and we provide conditions under which we can perform tractable inference. The experimental results confirm that practical learning is scalable to realistic datasets using this approach.
Cite
Text
Marchand et al. "Multilabel Structured Output Learning with Random Spanning Trees of Max-Margin Markov Networks." Neural Information Processing Systems, 2014.Markdown
[Marchand et al. "Multilabel Structured Output Learning with Random Spanning Trees of Max-Margin Markov Networks." Neural Information Processing Systems, 2014.](https://mlanthology.org/neurips/2014/marchand2014neurips-multilabel/)BibTeX
@inproceedings{marchand2014neurips-multilabel,
title = {{Multilabel Structured Output Learning with Random Spanning Trees of Max-Margin Markov Networks}},
author = {Marchand, Mario and Su, Hongyu and Morvant, Emilie and Rousu, Juho and Shawe-Taylor, John S},
booktitle = {Neural Information Processing Systems},
year = {2014},
pages = {873-881},
url = {https://mlanthology.org/neurips/2014/marchand2014neurips-multilabel/}
}