Aplikasi Metode Cross Entropy untuk Support Vector Machines

Budi Santosa, Tiananda Widyarini


Abstract


Support vector machines (SVM) is a robust method for  classification problem. In the original formulation, the dual form of SVM must be solved by a quadratic programming in order to get the optimal solution. The shortcoming of the standard version is as the classification problem is getting larger, the high computing time is needed. Cross entropy (CE) is a newly discovered optimization method with two main procedures: generating data samples by a chosen distribution, and updating the distribution parameters due to elite samples to generate a better sample in the next iteration. The CE method has been applied in many optimization problems with satisfying result. In this research, CE is applied to solve the optimization problem of Lagrangian SVM for faster  computational time. This method is tested in some real world datasets for classification problem. The results show that the application of CE in SVM is comparable to standard SVM in classifying two class data in terms of accuracy. In addition, this method can solve large datasets classification problem faster than standard SVM.

Keywords


Support vector machines, cross entropy, data mining, classification

Full Text:

pdf


DOI: https://doi.org/10.22219/JTIUMM.Vol10.No2.150-157 | Abstract views : 218 | pdf views : 218 |

Copyright (c) 2018 Jurnal Teknik Industri

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.


 
Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.