Feature selection for classification: A review (original) (raw)

The paper reviews feature selection techniques for classification, highlighting the challenges posed by high-dimensional, noisy data collected through modern technologies. It differentiates between feature extraction and feature selection, emphasizing the advantages of the latter in terms of interpretability and maintaining the original meaning of features. The review aims to consolidate existing literature on classification data and propose better methodologies for assessing feature relevance in the context of various classification tasks.