The tutorial introduces Hybrid Neural Networks (HyNNs), a novel architecture combining Convolutional Neural Network (CNN) and Multi-Layer Perceptron (MLP) branches for synthetic images and structured data, respectively. Participants will explore methods to transform tabular data into synthetic images, followed by an introduction to the TINTOlib library for real-world machine learning projects. Practical sessions focus on implementing HyNNs in regression and classification scenarios, emphasising hands-on learning.
The tutorial will conclude with a hands-on challenge where teams will compete to develop the most robust HyNN model using provided datasets. This competitive exercise allows participants to apply their knowledge to real-world scenarios, exploring the impact of image transformation methods on neural architectures. The winning team, determined by the lowest error rate, will showcase effective utilisation of HyNN models in this challenging and engaging environment.
Manuel Castillo-Cara (mamuelcastillo@dia.uned.es) is professor in the AI dept. at Universidad Nacional de Educación a Distancia (UNED). He holds a Ph.D. in Computer Science from Universidad de Castilla-La Mancha (UCLM). He develops the TINTOlib library and implements the method for converting tidy data into synthetic images called TINTO. He has given several seminars on the use of TINTOlib in Peru and Spain. He was a postdoctoral researcher in the AI dept. at the UPM, where he focuses on developing TINTOlib. He currently teaches courses on Machine Learning, Data Mining, and Deep Learning. He has over 20 publications in indexed journals, 1 patent, 2 software registrations, and has executed more than 15 national and international research projects, all focused on the area of AI and distributed computing.
Jiayun Liu (jiayun.liu@alumnos.upm.es)4 is Ph.D. student in the AI dept. at Universidad Politécnica de Madrid (UPM). Beginning his academic path with a Bachelor’s degree in Computer Science, Jiayun Liu furthered his knowledge by earning a Master’s in Data Science. Now in his initial year of Ph.D. research, he is extensively engaged in the field of Natural Language Processing, with a particular emphasis on large language models, graph neural networks, multimodal AI systems, and the interpretability of these complex models.
Raúl Garcı́a-Castro (r.garcia@upm.es) is Associate Professor in the Artificial Intelligence Department at Universidad Politécnica de Madrid (UPM). Having spent three years working as a software engineer, since he graduated in Computer Science in 2003 he has been working at UPM in the Ontology Engineering Group in more than twenty European and Spanish research projects, being Principal Investigator in six of them. He develops the TINTOlib library and implements the method for converting tidy data into synthetic images called TINTO. His research focuses on ontological engineering, ontology-based application integration, semantic interoperability, and Hybrid Neural Networks. He has authored more than 150 publications and regularly participates in standardization and in the conferences and workshops that are most relevant in his field.
