An efficient implicit Lagrangian twin bounded support vector machine

In this paper, an efficient implicit Lagrangian twin bounded support vector machine based on fuzzy membership is proposed with the dual formulation in order to reduce the sensitivity of noise and outliers. Here, the fuzzy membership values are determined according to distribution of the samples. We adopt the quadric and centroid fuzzy-based approach for LTBSVM and propose quadric based fuzzy membership approach and centroid based fuzzy membership approach for LTBSVM. The problems make strongly convex by using L2-norm of the vector of slack variable. Also, the solution of the problem is obtained through simple linear convergent iterative approach. Further, comparative performance analysis of proposed approach with state of art approaches have been done on standard real life with artificial datasets. This analysis announces that proposed approaches are effective in terms of generalisation performance and computational speed to other approaches. Our proposed approaches statistically validate and verify based on various parameters.