Journal Title : International Journal of Modern Trends in Engineering and Science

Author’s Name : N Anitha | S Deepa

Volume 02 Issue 02  Year 2015 

ISSN no: 2348-3121

Page no: 11-15

AbstractSupervised feature selection algorithms require a large amount of labeled training data. As a result, such algorithms provide insufficient information about the structure of the target concept, and can thus fail to identify the relevant features that are discriminative to different classes. On the other hand, unsupervised feature selection algorithms ignore label information and thus may lead to performance deterioration. For all these reasons, the usefulness of semi-supervised feature selection is more adapted and its effectiveness has been demonstrated. To overcome these drawbacks, we are proposing the semi supervised feature selection. In this work, proposed a framework for feature selection based on Partitioning the datasets, Constraint selection, Relevance of Features and redundancy elimination for semi-supervised dimensionality reduction. A new score function was developed to evaluate the relevance of features based on the locally geometrical structure of unlabeled data and the constraint preserving ability of labeled data.

Keywords— Supervised, Unsupervised, Semi supervised, Constraint, Relevance, Redundancy


[1] Z. Zhao and H. Liu, Spectral Feature Selection for Data Mining (Data Mining and Knowledge Discovery Series). Boca Raton, FL, USA: Chapman and Hall-CRC, (2012).
[2] S. T. Roweis and L. K. Saul, “Nonlinear dimensionality reduction by local linear embedding,” Science, Vol. 290, No. 5500, pp. 2323–2326, Dec.(2000).
[3] B. Scholkopf, A. Smola, and K. R. Muller, “Nonlinear component analysis as a Kernel Eigenvalue problem,” Neural Comput., vol. 10, no. 5, pp. 1299–1319, (1998).
[4] X. He and P. Niyogi, “Locality preserving projections,” in Proc. NIPS, (2004).
[5] M. Belkin and P. Niyogi, “Laplacian eigenmaps and spectral techniques for embedding and clustering,” in Proc. NIPS,(2002).
[6] I. Guyon and A. Elisseeff, “An introduction to variable and fea-ture selection,” J. Mach. Learn. Res., Vol. 3, pp. 1157– 1182, Mar.(2003).
[7] K. Benabdeslem and M. Hindawi, “Constrained Laplacian score for semi-supervised feature selection,” in Proc. ECMLPKDD, Athens, Greece, 2011, pp. 204–218.
[8] He.X, Cai.D, Yan.S and jiang Zhang.H,(2005)” Neighborhood Preservingembedding” ,in Proc.10th IEEE Int.Conf.Germany,pp.1208-1213.
[9] He.X and Niyogi .P, (2004)“Locality preserving projections,” in Proc. NIPS.

Full Paper: Click Here


Scroll Up