Machine Learning for Basic Visual Research in Graphic Design

This paper explores the intersection of machine learning and graphic design, aiming to enhance visual analysis methodologies through the integration of domain-specific knowledge. A critical examination of existing machine learning approaches for visual analysis reveals their limitations and the need to integrate more design specific knowledge. The paper proposes two approaches to analyze spatial aspects of graphic design. The application of the proposed methods demonstrates the potential of machine learning to reconstruct the intuition of graphic designers and to automate visual analysis tasks. The relevance of this study lies in its contribution to both academic research and practical applications in graphic design. By bridging the gap between computational methods and design theory, the study offers new perspectives on visual communication and provides tools for designers and researchers alike. Moving forward, interdisciplinary collaborations between machine learning experts and graphic designers will be essential to refine methodologies and unlock the full potential of machine learning in visual communication.