Kombinasi Metode 2DPCA (Two-Dimensional Principal Component Analysis), sPCA (sparse Principal Component Analysis), dan Ridge Regression Model dalam Pengenalan Wajah

Abstract

The 2DPCA (Two-Dimensional Principal Component Analysis) method, an extension of the PCA (Principal Component Analysis) method, is commonly used in pattern recognition to represent complex data in lower dimensions. One of its applications is in face recognition, where it encounters the challenge of ensuring good performance in different shooting conditions while avoiding overfitting to allow for good generalizations. However, it still faces the issue of overfitting. To overcome this problem, the authors researched by combining the 2DPCA method with the sPCA (sparse Principal Component Analysis) and the Ridge Regression Model. In this research, the 2DPCA method performs feature extraction, sPCA selects the most informative features or feature selection, and the Ridge Regression Model performs regularization. The results demonstrate that combining these three methods in face recognition can overcome overfitting and provide better accuracy than using the 2DPCA conventional, with an accuracy rate of 99.38%.