Francesca Lucchini and Camila Vera, both students of the IMFD and Master of Science in Engineering, mention in Computer Science in the Department of Computer Science of the P. Universidad Católica de Chile, (DCC UC) are currently focusing their research on clustering, that is, to create groups according to certain characteristics in the city of Santiago according to socioeconomic data, land use and urban visual perception.
In their work “Learning to clusterize urban areas: two competitive approaches and a user study”, they analyze and compare two methodologies: Graph Neural Networks and Gaussian Mixture Models, using graph networks and diffusion models for clustering, and a web survey to validate the results. .
The research, carried out together with IMFD researchers Hans Lobel, academic at DCC UC and the Department of Transportation Engineering and Logistics of the P. Universidad Católica, and Marcelo Mendoza, Department of Computer Science. P. Universidad Católica de Chile, seeks to put together a methodology that is capable of clustering the city, and at the same time, forming a spatial correlation, in order to detect neighborhoods.
“I really like working with graph networks, it seems to me an interesting area of study. I am also interested in applied research in the city, which can allow an updated vision, which helps to have more effective legislation”, explains Francesca Lucchini. “Urban computing is an area of study that allows the analysis of large volumes and various types of data that are constantly generated in cities. Our study discovers urban structures (such as neighborhoods) from different types of data using deep learning”, adds Camila Vera.