- Book Chapter
16
- 10.1007/978-981-13-1280-9_40
Lagrangian Twin-Bounded Support Vector Machine Based on L2-Norm
- Sep 12, 2018
- Umesh Gupta + 1 more +1
In this paper, a new convergent, flexible and easy technique is proposed for twin-bounded support vector machine (TBSVM) with the dual formulation. Here, a strongly convex objective function is constructed for proposed Lagrangian twin bounded support vector machine (LTBSVM) in consideration with L2-norm of the vector of slack variables in place of L1-norm vector of the slack variable. Twin support vector machine (TSVM) and TBSVM are used to solve quadratic programming problem (QPP) but in this proposed scheme, we consider an iterative method to solve a pair of linearly convergent equations. Further, a comparative analysis of generalized performance for TSVM, TBSVM, and proposed LTBSVM are tabulated for real world and as well as for synthetic datasets in consideration both the cases linear and nonlinear. This comparative analysis signifies the genuine improvement for the proposed method in terms of generalization performance and learning speed.
Read more