Review of Feature Selection Methods and Semi Supervised Feature Selection Algorithms for Classification

Current evolution in technology has resulted in tremendous growth of data with respect to dimensionality and sample size. Knowledge discovery from these high dimensional data are becoming cumbersome. Though powerful machine learning algorithms are prevalent, noisy data, imperfect sensing and collection of data leads to poor and defective knowledge discovery. Noise and redundant features cannot be circumvented due to which data collection process is biased. Dimensionality reduction techniques namely feature extraction and feature selections are the two most popular techniques to remove redundant and irrelevant features in huge dimensional data. Compared to feature extraction, feature selection leads to better readability and interpretability of features. This paper presents an extensive preliminary understanding about feature selection and attempts a comprehensive recent survey of semi supervised feature selection methods. This summary enables the researcher to obtain a deep understanding in choice of semi supervised feature selection algorithms for improving the learning performance. Keywords: Feature Selection, Semi supervised learning, Filters, Wrappers, Hybrid methods. Cite this Article: A. Meena Kowshalya, M.Lincy, R. Suvarna. Review of Feature Selection Methods and Semi Supervised Feature Selection Algorithms for Classification. International Journal of Software Computing and Testing. 2020; 6(1): 39–51p.