الفهرس 
introductionOnly 14 pages are availabe for public view
 |  | from | 115 |  |  |
| | | from | 115 | | |
Abstract
This thesis aims at improving the performance of some conventional classifiers by improving the reduction quality based on RST. Two models were introduced: Neighborhood Rough Bayesian model (NRB) and the Maximal Limited Similarity relation based Rough Set Model (MLS-RSM). To validate our models, we compare the classification accuracy of support vector machine, naive Bayes and random forests classifiers trained by our models and the related ones. With regard to classification accuracy, our models are more reliable in locating the feature subset than other RST extensions. At the same time, our models can achieve better classification accuracy than most of the unreduced datasets.