Face Recognition Approach using an Enhanced Particle Swarm Optimization and Support Vector Machine
Face recognition is one of the most promising research area in the last decades. The SVM approach is one of the famous approaches in machine learning fields because it can determine the global optimum solutions with lesser number of training samples especially, complex non-linear challenges such as...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Medwell Journals
2019
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/25612/1/Face%20Recognition%20Approach%20using%20an%20Enhanced%20Particle.pdf http://umpir.ump.edu.my/id/eprint/25612/ http://medwelljournals.com/abstract/?doi=jeasci.2019.2982.2987 http://dx.doi.org/10.3923/jeasci.2019.2982.2987 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Face recognition is one of the most promising research area in the last decades. The SVM approach is one of the famous approaches in machine learning fields because it can determine the global optimum solutions with lesser number of training samples especially, complex non-linear challenges such as in face recognition applications. Though, there is an important issue that can affects the whole classification process which is picking the optimum parameters of SVM. Recently, Particle Swarm Optimization (PSO) is used to discover the optimal parameters of SVM and many versions of PSO are used for this purpose, like: PSO-SVM technique, opposition PSO and SVM which called (OPSO-SVM) technique and AAPSO-SVM technique which represents adaptive acceleration PSO and SVM. In this study, a new hybrid technique based on the combination of "Accelerated PSO" and "OPSO-SVM" is introduced for face recognition applications. The hybridization can improve the convergence speed in PSO in order to find the optimal parameters of SVM. In the feature extraction process, the PCA algorithm is used for that purpose and the resulted features are delivered to the proposed technique in order to classify the face images. Two human face datasets are used in the experimentation stage such as, SCface dataset and CASIA face dataset in order to validate the performance of the proposed technique. The comparison process for proposed technique with the other recent technique, like: PSO-SVM, OPSO-SVM and AAPSO-SVM is done as an assessment process. The proposed technique provided high accuracy for recognition when we compared it with the other techniques and it was robust in finding the optimal parameters of SVM. |
|---|