Fuzzy possibilistic on different search spaces
Hossein Yazdani · 2016
Extracting knowledge from data is a sophisticated procedure that needs to be addressed by a comprehensive method. To get the accurate information from each repository some learning methods have been introduced, and just some of them are able to provide the most accurate results. Fuzzy, probability, possibilistic, and bounded fuzzy possibilistic are some of the most common partitioning methods that prepare the most flexible environment for data objects. The paper compares the accuracy of these methods on some data sets. The paper also introduces some algorithms to cover diversity in feature spaces besides vector space. The introduced algorithms are implemented based on bounded fuzzy possibilistic methods, in order to be compared with conventional fuzzy and possibilisitc methods. Learning methods are also compared on their membership assignments when using Euclidean or Weighted Distance Function (WFD). The results show that the introduced algorithms perform better than the other conventional methods on data sets presented in this literature. Results also show that the methods with weighted distance function in their similarity functions are capable of covering diversity in search spaces.