ML Anthology

Improvements to Platt's SMO Algorithm for SVM Classifier Design

S. Sathiya Keerthi, Shirish K. Shevade, Chiranjib Bhattacharyya, K. R. K. Murthy
NeCo 2001 pp. 637-649
doi:10.1162/089976601300014493 /neco/2001/keerthi2001neco-improvements/

Abstract

This article points out an important source of inefficiency in Platt's sequential minimal optimization (SMO) algorithm that is caused by the use of a single threshold value. Using clues from the KKT conditions for the dual problem, two threshold parameters are employed to derive modifications of SMO. These modified algorithms perform significantly faster than the original SMO on all benchmark data sets tried.

NeCo Semantic Scholar

Cite

Text

Keerthi et al. "Improvements to Platt's SMO Algorithm for SVM Classifier Design." Neural Computation, 2001. doi:10.1162/089976601300014493

Markdown

[Keerthi et al. "Improvements to Platt's SMO Algorithm for SVM Classifier Design." Neural Computation, 2001.](https://mlanthology.org/neco/2001/keerthi2001neco-improvements/) doi:10.1162/089976601300014493

BibTeX

@article{keerthi2001neco-improvements,
  title     = {{Improvements to Platt's SMO Algorithm for SVM Classifier Design}},
  author    = {Keerthi, S. Sathiya and Shevade, Shirish K. and Bhattacharyya, Chiranjib and Murthy, K. R. K.},
  journal   = {Neural Computation},
  year      = {2001},
  pages     = {637-649},
  doi       = {10.1162/089976601300014493},
  volume    = {13},
  url       = {https://mlanthology.org/neco/2001/keerthi2001neco-improvements/}
}