Semantic mapping though neural networks: the selforganizing maps (som) as representation of patterns and fields

• Main Authors: Francisco Javier Abarca-Alvarez, Fernando Osuna Pérez
• Format: Article
• Language: Spanish
• Published: Universitat Politècnica de València 2013-11-01
• Series: EGA
• Subjects: cartografía semántica; mapa auto-organizado; som; patrón; estructura
• Online Access: https://polipapers.upv.es/index.php/EGA/article/view/1692

The Science of Artificial Intelligence provides us with techniques to improve our understanding and characterization of the coherences and patterns which constitute reality. Among these, artificial neural networks and more specifically Self Organizing Maps (SOM) stand out because of their ability to map reality in such a way that their objectives are represented distributed and structured two-dimensionally, with their properties as a single starting point. In this way an entire series of topological relations is generated, which in their turn enable the grouping and characterization of reality. In this research these representations are explored as a valid method to obtain information and to interpret reality. By means of experimentation this kind of methods are implemented to further understanding of diverse exemplary residential fabrics, while obtaining a typological grouping which enables the characterization of urban forms starting from their defining variables.