Programming for Symbolic Model Discovery
Current Neural Network based approaches to AI, such as LLMs and CNNs, have produced incredible results in recent years. This has been achieved in a wide variety of domains, that include NLP and computer vision, to mention but a couple of prominent examples. However, while Neural Networks have proven to be powerful learning tools, they intrinsically rely on sub-symbolic processing of data (pattern matching) and in general do not exploit symbolic modelling or knowledge representation, which some have argued are the reason for their current limitations. Recently, some AI systems have been developed that incorporate evolutionary search to enhance the performance of LLMs, with impressive results, such as those reported by AlphaEvolve, FunSearch and . These systems are large scale deployments of what is essentially a Genetic Programming (GP) search, which uses evolutionary principles to search for models that are constructed and manipulated symbolically. The models produced by GP have the intrinsic potential of being interpretable, given their symbolic representation, which is a desirable property of machine learning models that will be used in real-world scenarios. This talk will provide a brief introduction to Genetic Programming, how it is currently being used and developed, and why it can be a promising tool in the development of new AI systems.
Resumen curricular
Dr. Leonardo Trujillo is Professor at the Tecnológico Nacional de México/Instituto Tecnológico de Tijuana (ITT), working in the Department of Electrical and Electronic Engineering, and the Engineering Sciences Graduate Program. Dr. Trujillo received an Electronic Engineering degree and a Masters in Computer Science from ITT, as well as a doctorate in Computer Science from CICESE research center in Ensenada, Mexico. His main research interests include evolutionary computation, genetic programming, machine learning and artificial intelligence. Dr. Trujillo has been the PI of several national and international research grants, research that has been extensively published in a variety of journals, conferences and edited books. He is currently Editor-in-Chief of the Genetics Programming and Evolvable Machines journal (Springer), associate editor of the European Journal of Artificial Intelligence (Sage) and Mathematical and Computational Applications (MDPI), and series co-chair of the NEO Workshop series, and has previously been co-organizer of the Genetic Programming Theory and Practice Workshop series. He is also General Chair of the ACM-SIGEVO Genetic and Evolutionary Computation Conference (GECCO 2026).
NOTA. Todos los seminarios se graban y están a tu disposición en el Canal de YouTube del Departamento, consúltalos:
https://www.youtube.com/channel/UCwZp9bGRHC1FT5lof7kuLJg/playlists
Lugar: Auditorio Depto. Ciencias de la Computación y en línea https://us02web.zoom.us/j/86823883261?pwd=77I8m6bBJTulZtvYCQB8VMkg43uDWx.1
Fecha: 30-01-2026
Hora: 12:00 pm
