The World Wide Web is growing rapidly and the Internet users are still increasing day by day. Increasing with the number of users, the need for automatic classification techniques with good classification accuracy increases as search engines depend on previously classified web pages stored as classified directories to retrieve the relevant results. Machine learning techniques for automatic classification gains more interest as the classifier improves its performance with experience. In this paper we propose a method called Combined Feature Selection and Classification for effective categorization of web pages. Our experimental results show that our proposed approach improves the classification accuracy with the optimum number of attributes. We experimented with four machine learning classifiers (CV Parameter Selection, Logit Boost, Random Committee and VFI).Our results effectively improve the accuracy.

In this study, we use a recursive method based upon power series to solve nonlinear Volterra integral equations system of the second kind. This method gives an approximate solution as the Taylor expansion for the solution of the system via some simple computations. Numerical examples illustrate the pertinent features of the method.

TV minimizing based PDE models have been proved to be the most effective tool for image restoration. However for many applications, it is desirable to reduce or remove the staircasing effect of a TV solution to restore more realistic contrast, features, geometry and textures. In this paper, we propose and investing a bounded constrained regularization technique to improve the TV solution. A numerical method about the partial differential equations is attempted. Notable improvements are obtained by numerical experiments. AMS subject classification: 68U10, 65K10

Bounds for the extreme eigenvalues involving trace and determinant are presented. Also, we give the upper bounds for the Perron root of a nonnegative symmetric matrix under certain conditions.

In this paper, we study the initial value problem of the Generalized KdV equation, define a )3s ≥ . nonlinear map Therefore, the solution of the Generalized KdV equation with arbitrary precision on Turing machines can be satisfied.

In this paper, the nonclassical symmetries and group invariant solutions of the Boussinesq- Burgers equation have been discussed. By using the nonclassical method, we obtain nonclassical symmetries that reduce the Boussinesq-Burgers equation to ordinary differential equation, and several invariant solutions. We remark that some of them are new solutions of the Boussinesq-Burgers equation.

Based on the Chen-Harker-Kanzow-Smale smoothing function, a smoothing method with a smoothing variable is proposed for solving the second-order cone programming. Our algorithm needs to solve only one linear system of equations and to perform only one line search at each iteration. Without restrictions regarding its starting point, the algorithm is shown to be globally convergent.

This paper presents a fuzzy optimization method based on the inclusion degrees of intuitionistic fuzzy sets to solve multi-criteria decision making problems under fuzzy environments. First, the inclusion degree of intuitionistic fuzzy sets is defined and a series of specific formulas of inclusion degree are presented by means of the normal implication operators. Some formulas of inclusion degree of intuitionistic fuzzy sets are generalized by defining the cardinal number of intuitionistic fuzzy sets. Then, we give multi- criteria fuzzy decision-making method based on inclusion degree of intuitionistic fuzzy sets. Finally, we illustrate the effectiveness of the method proposed in this paper by an example.