Fernando Sancho

Since the beginning of my research life I have been interested in using mathematical tools as a means to model complex systems, which for the moment are resistant to being modelled by traditional means. In particular, I am interested in solving problems that require high computational resources to solve. Thus, my doctoral thesis was focused on the use of unconventional bio-inspired systems (mainly based on the use of DNA molecules or cells as basic units of calculation). Subsequently, and motivated by the type of limitations we found in modelling with these means, I focused on the mathematical study of complex networks as universal representation systems for highly complex systems and complex data, as well as the mathematical analysis and control of multi-agent systems formed by especially simple units (such as cellular automata).
In recent years, and always with the aim of better understanding complex systems through mathematical modelling, I have begun to make use of Machine Learning techniques for the modelling, analysis, prediction and control of complex dynamic systems and data obtained from them, with particular emphasis on those that have a very high semantic load that makes it difficult to process them by means of the most common data analysis techniques. It is therefore necessary to extend the methodological tools of complex networks in order to introduce a layer of measures and procedures capable of manipulating semantic information, taking advantage of the long-distance relationships established between their units. As a result of the methodologies developed, we have presented results with mathematical applications to fields traditionally alien to their techniques, with special attention to the study of systems of cultural objects, as well as the prediction of medical dynamics and mechanical engineering systems by using Deep Learning techniques.