Adaptive learning evaluation model for evolutionary art

Evolutionary art system is facing the challenge of how aesthetic judgement can be automated for use as a fitness function. Judging beauty is a highly subjective task, but certain features are considered important in aesthetic judgement. This paper introduces an adaptive learning evaluation model for guiding the evolutionary process. Certain aesthetic features are extracted from internal evolutionary images and external real world paintings, which are then selected by the model. The model is built by selecting learning approach with better accuracy by training these features. Multi-layer perceptron and C4.5 decision tree are compared for machine learning of aesthetic judgements. Our results show that these features play important roles in aesthetic judgements and the adaptive model is efficient at predicting user´s preference.