Aesthetic Learning in an Interactive Evolutionary Art System

Abstract

Learning aesthetic judgements is essential for reducing the users’ fatigue in evolutionary art system. Although judging beauty is a highly subjective task, we consider that certain features are important to please users. In this paper, the aesthetic preferences are explored by learning the features, which we extracted from the images in the interactive generations. In addition to color ingredients, image complexity and image order are considered highly relevant to aesthetic measurement. We interpret these two features from the information theory and fractal compression perspective. Our experimental results suggest that these features play important roles in aesthetic judgements. Our findings also show that our evolutionary art system is efficient at predicting user’s preference.