Genetic Algorithm as Quadratic Programming Solver for Support Vector Machine

Nik Ahmad Akram, Niusha Shafiabady, Dino ISA


In this paper we discuss a new approach to solve Quadratic Programming problem in Support Vector Machine (SVM) usingGenetic Algorithm. A new strategy is introduced to reduce the number of generations while retaining the SVM accuracy. The approach istested using a real-time dataset.The proposed approach takes advantage of the multimodal optimization ability of Genetic Algorithm inaddition to the classificationcharacterization of SVM by including GA in SVM training phase. This approach incorporates the exploitationand exploration power of GA with generalization ability of SVM at the same time.The achieved results show that the proposed method hashad a good performance.

Full Text:



  • There are currently no refbacks.