Fast Linear SVM Validation Based on Early Stopping in Iterative Learning Mahmoud Famouri * , Mohammad Taheri † and Zohreh Azimifar ‡ Department of IT, Computer Science and Engineering School of Electrical and Computer Engineering Shiraz University, Shiraz, Iran * famouri@cse.shirazu.ac.ir † taheri@cse.shirazu.ac.ir ‡ azimifar@cse.shirazu.ac.ir Received 15 October 2013 Accepted 5 August 2015 Published 29 September 2015 Classi¯cation is an important ¯eld in machine learning and pattern recognition. Amongst various types of classi¯ers such as nearest neighbor, neural network and Bayesian classi¯ers, support vector machine (SVM) is known as a very powerful classi¯er. One of the advantages of SVM in comparison with the other methods, is its e±cient and adjustable generalization capability. The performance of SVM classi¯er depends on its parameters, specially regularization parameter C, that is usually selected by cross-validation. Despite its generalization, SVM su®ers from some limitations such as its considerable low speed training phase. Cross-validation is a very time consuming part of training phase, because for any candidate value of the parameter C, the entire process of training and validating must be repeated completely. In this paper, we propose a novel approach for early stopping of the SVM learning algo- rithm. The proposed early stopping occurs by integrating the validation part into the optimi- zation part of the SVM training without losing any generality or degrading performance of the classi¯er. Moreover, this method can be considered in conjunction with the other available accelerator methods since there is not any dependency between our proposed method and the other accelerator ones, thus no redundancy will happen. Our method was tested and veri¯ed on various UCI repository datasets and the results indicate that this method speeds up the learning phase of SVM without losing any generality or a®ecting the ¯nal model of classi¯er. Keywords : Support vector machine; cross-validation; early stopping; generalization; parameter selection. 1. Introduction Classi¯cation plays an important role in machine learning and pattern recognition. In such a problem, each sample, with respect to its features, belongs to a proper class. ‡ Corresponding author. International Journal of Pattern Recognition and Arti¯cial Intelligence Vol. 29, No. 8 (2015) 1551013 (20 pages) # . c World Scienti¯c Publishing Company DOI: 10.1142/S0218001415510131 1551013-1 Int. J. Patt. Recogn. Artif. Intell. Downloaded from www.worldscientific.com by Mr. Mahmoud Famouri on 10/07/15. For personal use only.