A Study on Techniques for Online Feature Choice Method

International Journal of Computer Science and Engineering

© 2015 by SSRG - IJCSE Journal

Volume 2 Issue 3

Year of Publication : 2015

Authors : B.Ajay Babu, Nimala.K

10.14445/23488387/IJCSE-V2I3P112

How to Cite?

B.Ajay Babu, Nimala.K, "A Study on Techniques for Online Feature Choice Method," SSRG International Journal of Computer Science and Engineering , vol. 2,  no. 3, pp. 1-5, 2015. Crossref, https://doi.org/10.14445/23488387/IJCSE-V2I3P112

Abstract:

Feature choice is a vital technique for data processing. Despite its importance, most studies of feature choice are restricted to batch learning. in contrast to ancient batch learning ways, on-line learning represents a promising family of economical and ascendable machine learning algorithms for large-scale applications. Most existing studies of on-line learning need accessing all the attributes/features of coaching instances. Such a classical setting isn't invariably acceptable for realworld applications once data instances square measure of high spatial property or it's overpriced to accumulate the total set of attributes/features. to deal with this limitation, we investigate the matter of on-line Feature choice (OFC) during which an internet learner is simply allowed to keep up a classifier concerned only atiny low and glued range of options. The key challenge of on-line Feature choice is a way to create correct prediction for associate degree instance employing a tiny range of active options. will be in distinction to the classical setup of on-line learning wherever all the options can be used for prediction. we tend to decide to tackle this challenge by finding out meagerness regularization and truncation techniques. Specifically, this article addresses 2 totally different tasks of on-line feature selection: (1) learning with full input wherever associate degree learner is allowed to access all the options to make a decision the set of active options, and (2) learning with partial input wherever solely a restricted range of options is allowed to be accessed for every instance by the learner. we tend to gift novel algorithms to unravel every of the 2 issues and provides their performance analysis. we tend to value the performance of the projected algorithms for on-line feature choice on many public data base, and demonstrate their applications to realworld issues as well as image classification in laptop vision and microarray gene expression analysis in bioinformatics. The encouraging results of our experiments validate the effectualness and potency of the proposed techniques.

Keywords:

Feature Choice; on-line Learning; Large-scale knowledge Mining; Classification

References:

[1] R. Bekkerman, R. El-Yaniv, N. Tishby, and Y.Winter. Distributional word clusters vs. words for text categorization. Journal of Machine Learning Research, 3:1183–1208, 2003. 
[2] J. Bi, K. P. Bennett,M. J. Embrechts, C. M. Breneman, and M. Song. Dimensionality reduction via sparse support vector machines. Journal of Machine Learning Research, 3:1229– 1243, 2003. 
[3] G. Cavallanti, N. Cesa-Bianchi, and C. Gentile. Tracking the best hyperplane with a simple budget perceptron. Machine Learning, 69(2-3):143–167, 2007. 
[4] N. Cesa-Bianchi, S. Shalev-Shwartz, and O. Shamir. Efficient learning with partially observed attributes. Journal of Machine Learning Research, pages 2857–2878, 2011. 
[5] A. B. Chan, N. Vasconcelos, and G. R. G. Lanckriet. Direct convex relaxations of sparse svm. In ICML, pages 145–153, 2007. 
[6] K. Crammer, O. Dekel, J. Keshet, S. Shalev-Shwartz, and Y. Singer. Online passive-aggressive algorithms. J. Mach. Learn. Res. (JMLR), 7:551–585, 2006. 
[7] K. Crammer, M. Dredze, and F. Pereira. Exact convex confidenceweighted learning. In NIPS, pages 345–352, 2008. [8] M. Dash and H. Liu. Feature selection for classification. Intell. Data Anal., 1(1-4):131–156, 1997. 
[9] O. Dekel, S. Shalev-Shwartz, and Y. Singer. The forgetron: A kernel-based perceptron on a budget. SIAM J. Comput., 37(5):1342–1372, 2008. 
[10] C. H. Q. Ding and H. Peng. Minimum redundancy feature selection from microarray gene expression data. J. Bioinformatics and Computational Biology, 3(2):185–206, 2005.