Toward Applying Machine Learning to Design Rule Acquisition for Automated Graphics Generation

Effective automated graphics generation systems rely on well-established design rules. We are currently exploring how machine learning techniques may be used to acquire design rules automatically. In this paper, we present a model that addresses four fundamental aspects toward applying machine learning to graphics design rule formulation. In particular, learning spaces and learning goals describe where learning may take place and what needs to be learned. Learning features and learning strategies discuss how to describe various learning data and how to choose proper learning techniques. Using our model, we have designed and conducted two experiments to demonstrate two different applications of machine learning in graphics rule acquisition.