We show how to train SVMs with an optimal guarantee on the number of support vectors (up to constants), and with sample complexity and training runtime bounds matching the best known for kernel SVM optimization (i.e. without any additional asymptotic cost beyond standard SVM training). Our method is simple to implement and works well in practice.
|Original language||American English|
|Number of pages||9|
|State||Published - 2013|
|Event||30th International Conference on Machine Learning, ICML 2013 - Atlanta, GA, United States|
Duration: 16 Jun 2013 → 21 Jun 2013
|Conference||30th International Conference on Machine Learning, ICML 2013|
|Period||16/06/13 → 21/06/13|