Error bounds for support vector machines with application to the identification of active constraints

De Leone, Renato; Lazzari, C.

In this paper, we present a new global error bound for the dual formulation of the support vector machine (SVM) whose penalty terms are measured by 2-norm (L2-SVM). An further formulation in which an additional term is added to the objective function is also considered (SCL2-SVM). A global error bound for the dual of the latter problem is derived. This error bound is used to identify those constraints of primal SCL2-SVM that will be active at the solution. The correct identification of active constraints is important to find the support vectors and to build the working set of most iterative algorithms for SVM. Computational results are presented showing that the proposed method is able to quickly identify a large percentage of constraints that are active at the optimal solution.

Scheda prodotto non validato

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Titolo:	Error bounds for support vector machines with application to the identification of active constraints
Autori:	DE LEONE, Renato LAZZARI C. mostra contributor esterni nascondi contributor esterni
Data di pubblicazione:	2010
Rivista:	OPTIMIZATION METHODS & SOFTWARE
Abstract:	In this paper, we present a new global error bound for the dual formulation of the support vector machine (SVM) whose penalty terms are measured by 2-norm (L2-SVM). An further formulation in which an additional term is added to the objective function is also considered (SCL2-SVM). A global error bound for the dual of the latter problem is derived. This error bound is used to identify those constraints of primal SCL2-SVM that will be active at the solution. The correct identification of active constraints is important to find the support vectors and to build the working set of most iterative algorithms for SVM. Computational results are presented showing that the proposed method is able to quickly identify a large percentage of constraints that are active at the optimal solution.
Handle:	http://hdl.handle.net/11581/9999
Appare nelle tipologie:	Articolo

File in questo prodotto:

Non ci sono file associati a questo prodotto.

Utilizza questo identificativo per citare o creare un link a questo documento:
http://hdl.handle.net/11581/9999

Citazioni

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.