Research On Classification Problems Based On The Information Entropy Of The Rough Sets's Reduction And Support Vector Machine

With the development of science and technology, the information from all the fields is increasing rapidly. It has been to a big problem that how to acquire the useful knowledge and the effective classification methods from these data. The rough set method, which is an excellent data analysis tool to process the uncertain information such as imprecise, inconsistent, incomplete and so on. It has been successfully applied to machine learning and pattern recognition, decision support and knowledge discovery, fault diagnosis and etc. Knowledge reduction of rough set as the key technology which can process data rapidly and effectively.Support vector machine is a new machine learning technique developed from the middle of 1990s by Vapnik. It's characterized by the use of a maximal margin hyperplane, the theory of kernels and the absence of local minima, convex optimization the sparseness of the solution, Mercer's theorem and the capacity control obtained by acting on the margin. And it's a new tool for machine learning by using optimization method. Because support vector machines has not only simpler structure, but also better performance, especially its better generalization ability. In recent years, it has been widely applied to classification and pattern recognition, etc. Since many multiclass problems can be divided into binary classification problems, the original problem of support vector machine is designed to slove the binary classification.Through research the theory above and use their advantages, we designed a method which is base on the information entropy of the rough sets's reduction and support vector machine. Since the algorithm is applied to the classification system,the result demonstrates that the accuracy and speed of classification are improved.The main work is given as follows:1. An overview on a variety of algorithms and techniques for the reduction method of rough set, and select the information entropy as the method of reduction algorithm.2. Introduce a new reduction method which is base on the discernibility matrix.3. The paper designs a new kind of classification algorithm which is base on rough set theory and support vector machine method. We preprocess the SVM input training data by applying feature selection which is conducted by attribute reduction,so the dimension of the input vector is decrease. And using the reduction algorithm of this paper to remove redundant data and fix inconsistent information ,so the quality of hyperplane is optimized.4. After some brief research and experimentation were made on some of UCI data sets, and the results demonstrate the effectiveness of this algorithm.