-

 

 

 




Optimization Online





 

Detecting relevant variables and interactions for classification in Support Vector Machines

Emilio Carrizosa (ecarrizosa***at***us.es)
Belen Martin-Barragan (belmart***at***us.es)
Dolores Romero Morales (dolores.romero-morales***at***sbs.ox.ac.uk)

Abstract: The widely used Support Vector Machine (SVM) method has shown to yield good results in Supervised Classification problems. The Binarized SVM (BSVM) is a variant which is able to automatically detect which variables are, by themselves, most relevant for the classifier. In this work, we extend the BSVM introduced by the authors to a method that, apart from detecting the relevant variables, also detects the most relevant interactions between them. The classification ability of the proposed method is comparable to standard SVM for different kernels and clearly better than Classification Trees. Our method involves the optimization of a Linear Programming problem with a large number of decision variables, for which we use the well-known Column Generation technique.

Keywords: Data Mining, Supervised Classification, Column Generation, Support Vector Machines, Interactions Detection.

Category 1: Applications -- Science and Engineering (Data-Mining )

Citation:

Download: [PDF]

Entry Submitted: 05/05/2006
Entry Accepted: 05/05/2006
Entry Last Modified: 05/05/2006

Modify/Update this entry


  Visitors Authors More about us Links
  Subscribe, Unsubscribe
Digest Archive
Search, Browse the Repository

 

Submit
Update
Policies
Coordinator's Board
Classification Scheme
Credits
Give us feedback
Optimization Journals, Sites, Societies
Mathematical Programming Society