FUZZY SIMILARITY RELATION AND IT’S APPLICATION IN FEATURE SELECTION

Abstract

Owing to technology advancements and the rising expansion of electronically stored information, automated solutions are required to assist users in processing and maintaining this large volume of information. The primary sources of knowledge are subject matter experts and computer program that evaluate enormous amounts of data using machine learning. Knowledge extraction is a crucial process stage in the construction of clever and skilled systems. However, because of the noise and the volume of data, the knowledge extraction stage is extremely sluggish or perhaps impossible. The effectiveness of classifiers and the readability of data in machine learning algorithms both benefit from the decision of pertinent and characteristics without repetition. This process the term "feature selection" or attribute reduction. Numerous domains, such as the use of image processing, artificial intelligence, bioinformatics, data mining, natural language processing, etc., use feature selection in ways that are very relevant to expert and intelligent systems. The discretization process may result in some information being lost, rendering rough set theory unsuitable for attribute reduction of real-valued data sets, despite the fact that it has been employed effectively for attribute reduction. Real-valued data can be handled easily thanks to the numerous attribute selection algorithms that have been given, In addition, the integration of collection of blurry and rough theories.