<![CDATA[Face recognition using artificial intelligence has extensive usage. A facial image can be used as face-unlock on mobile devices, biometrics on attendance systems, and auto-tagging images on social media. However, the face is one of the most challenging objects to be modeled as it is affected by the age, lighting, location of image capture, orientation, pose, and expression. Face images can be decomposed to take out the main components (on low frequency) to be an identifier. The image decomposition process can be done using Wavelet transform. This study use Neural Probabilistic Network (PNN) method to classify the facial images based on Wavelet feature extraction. The aims of this study is to implement the PNN classification method and Wavelet feature extraction to build a facial image classification model. The wavelet decomposition levels used in the study are levels 2 to 6. Meanwhile, the K-Fold Cross Validation method is used to split the data into training data and test data. The total of facial images used is 800 images, consist of 40 individuals with 20 individual images per person. The facial image data was downloaded from the University of Essex, United Kingdom. This study showed that an enhancement accuracy along with the increased Wavelet decomposition levels from 2 to 5. The best accuracy was obtained using Wavelet decomposition level 5, which was 97.25%. Whereas on Wavelet decomposition level 6 the was an accuracy reduction of 1.88% to become 95.37%.]]>
