Enhanced Fuzzy Roughset based Feature Selection Strategy using Differential Evolution
No Thumbnail Available
Date
2013
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Dimensionality reduction technique aims to c J ^ ^ c e the performance o f the classification
model by removing the irrelevant and redfummm data. Reducing the number o f attributes in
the classification model helps to alleviate^Mp curse o f dimensionality, where it is the major
crisis to data storage and also facilitates the mining techniques like
classification, clustering, communi^lmb^, visualization and high-dimensional data storage.
Once the dimensionality o f a n g ^ J ^ increases then it leads the data to sparsity. Sparsity
problem is quite tricky for e^ytechnique that entails statistical importance. Thus, it is
necessary to remove the^ ir iy ^ a n t and redundant features, augment the performance o f the
classifier model, speed learning task, interpret the model and highlight the vital
features with their ralmidm. In this paper, a modfied feature selection technique based on
fuzzy Roughset t h e ^ ^ n d Differential Evolution is proposed. Here, the experimental results
are carried outtSmhg binary and multiclass datasets taken from UCI Machine Learning
R ep o sitory\