Decomposition Algorithms for Training Large-Scale Semiparametric Support Vector Machines

Abstract

We describe a method for solving large-scale semiparametric support vector machines (SVMs) for regression problems. Most of the approaches proposed to date for large-scale SVMs cannot accommodate the multiple equality constraints that appear in semiparametric problems. Our approach uses a decomposition framework, with a primal-dual algorithm to find an approximate saddle point for the min-max formulation of each subproblem. We compare our method with algorithms previously proposed for semiparametric SVMs, and show that it scales well as the number of training examples grows.

Cite

Text

Lee and Wright. "Decomposition Algorithms for Training Large-Scale Semiparametric Support Vector Machines." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2009. doi:10.1007/978-3-642-04174-7_1

Markdown

[Lee and Wright. "Decomposition Algorithms for Training Large-Scale Semiparametric Support Vector Machines." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2009.](https://mlanthology.org/ecmlpkdd/2009/lee2009ecmlpkdd-decomposition/) doi:10.1007/978-3-642-04174-7_1

BibTeX

@inproceedings{lee2009ecmlpkdd-decomposition,
  title     = {{Decomposition Algorithms for Training Large-Scale Semiparametric Support Vector Machines}},
  author    = {Lee, Sangkyun and Wright, Stephen J.},
  booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
  year      = {2009},
  pages     = {1-14},
  doi       = {10.1007/978-3-642-04174-7_1},
  url       = {https://mlanthology.org/ecmlpkdd/2009/lee2009ecmlpkdd-decomposition/}
}