Predicting Website's Visual Appeal using a Hybrid of Convolutional Neural Network and Support Vector Machine

Abstract

In predicting a website's visual appeal, hand-crafted feature extractors may not be able to determine all the relevant features to extract from a screenshot of a website's homepage. Also, since it is hand-crafted, there is a need to determine what features to extract. This study aims to use a hybrid of convolutional neural network and support vector machine to predict a website's visual appeal. Using the hybrid CNN-SVM model, the AI system would extract features from an image of a website's homepage and determine its visual appeal with respect to the age, gender, country, and educational level of the target users.