Antonio Liotta

Faculty of Engineering, Free University of Bozen-Bolzano, Bolzano, Italy

9:00-10:00

Making Data AI-Ready: why better data outperforms bigger models

Abstract: Advanced machine learning models rarely fail because of insufficient algorithmic sophistication, they fail because the data pipeline is not AI-ready. In large, heterogeneous, and imperfect datasets, bias, instability, and weak generalization are typically consequences of data design rather than model choice. This lecture presents a structured, data-centric methodology for transforming raw data into robust ML systems, emphasizing collection and representation, feature extraction, dimensionality reduction, bias mitigation, leakage prevention, and evaluation beyond accuracy. Using a music genre classification case study, we show how systematic data curation and iterative pipeline optimization can substantially improve generalization performance, often yielding higher return on investment than increasing model complexity. Participants will gain a practical framework for building scalable, reproducible, and resource-aware ML pipelines that extract maximal value from data before resorting to more complex models.

Short Bio: Antonio Liotta is Full Professor of Data Science and Machine Learning at the Faculty of Engineering, Free University of Bolzano (Italy), where he leads the Data-Driven Artificial Intelligence research area and co-directs the Master’s programme in Data Analytics for Economics and Management. He previously founded the Data Science Research Centre at the University of Derby. For more than three decades, his work has explored a central question: how can we transform complex, messy, real-world data into reliable and efficient intelligent systems? His research bridges data science, machine learning, and intelligent infrastructures, with applications spanning smart cities, Internet of Things, energy systems, artificial vision, and human-centred AI. He is widely recognized for pioneering contributions to micro-edge intelligence and sparse neural networks for embedded learning, advancing AI that is not only powerful, but scalable and sustainable. Professor Liotta has authored over 350 scientific publications and collaborates internationally across disciplines to advance data-centric AI. He is Editor-in-Chief of the Springer Internet of Things book series and co-author of the books Networks for Pervasive Services and Data Science and Internet of Things. His work continues to focus on building AI systems that create value from data in robust, responsible, and resource-aware ways.

Paolo Sorino

Department of Electrical and Information Engineering, Polytechnic University of Bari, Bari, Italy

10:00-10:45

Artificial Intelligence and Explainable AI in Clinical Decision Support Systems: Innovation, Interpretability, and Trust

Abstract: Artificial Intelligence and Explainable AI in Clinical Decision Support Systems: Innovation, Interpretability, and Trust Artificial Intelligence is rapidly transforming Clinical Decision Support Systems (CDSS), enabling new capabilities for prediction, diagnosis, personalization, clinical documentation, and workflow automation. At the same time, the increasing complexity of machine learning models and emerging agentic architectures raises critical questions regarding interpretability, safety, accountability, and clinical trust. This lecture provides an overview of the evolution of CDSS from rule-based systems to contemporary AI-driven and agent-oriented solutions. We discuss the main innovation patterns in clinical AI, including predictive models, multimodal learning, medical imaging, large language models, and AI agents. Particular attention is devoted to Explainable Artificial Intelligence (XAI) as a key mechanism for supporting transparent and trustworthy decision-making in healthcare settings. The lecture introduces practical explainability techniques, with a focus on SHAP-based feature attribution methods for clinical prediction models, illustrating how explanations can support model inspection, clinical validation, and human oversight. A short hands-on coding demonstration will show how explainability methods can be applied to a healthcare-oriented machine learning model and how their outputs can be interpreted in a clinical context. Finally, the session discusses trust, governance, regulatory frameworks, validation strategies, and human-AI collaboration, highlighting the challenges of deploying AI-enabled CDSS in real-world healthcare environments. The central message is that clinical AI should be evaluated not only as a predictive model but as a socio-technical system where performance, explainability, workflow integration, and accountability jointly determine its clinical value.

Short Bio: Paolo Sorino is a Post-Doctoral Researcher in Information Processing Systems (ING-INF/05) at the Polytechnic University of Bari. He received a PhD in Information Engineering from Politecnico di Bari, defending a thesis entitled “Leveraging Artificial Intelligence for Enhanced and Human-centred Healthcare Solutions”. His research focuses on the integration of Artificial Intelligence to develop innovative and human-centred healthcare solutions, as well as on the analysis and modeling of biological signals to improve the effectiveness and accessibility of healthcare services. His scientific interests include Machine Learning and Deep Learning for clinical prediction, Explainable AI (XAI), Brain-Computer Interfaces (BCI), and biomedical signal analysis (EEG/ECG). He is the author of more than 30 peer-reviewed publications and actively participates in national and European research projects on AI-driven healthcare systems.

Coffee Break

10:45-11:00

Sergio Miranda

Department of Humanities, Philosophy and Education, University of Salerno, Italy

11:00-12:00

Artificial Intelligence in Education

Abstract: This talk traces the epistemological and technological evolution from classical Information Theory and early Artificial Neural Networks to contemporary Deep Learning and Generative AI paradigms. The work highlights key socio-technical milestones in Technology-Enhanced Learning (TEL), including the engineering of e-learning platform and the subsequent implementation of Learning Analytics. A central focus is dedicated to advanced cognitive scaffolding mechanisms. Additionally, the incorporation of AI-generated virtual speakers is examined as a method to enhance interactive, multilingual e-learning experiences, alongside evidence-informed AI frameworks aimed at refining teacher training and fostering professional expertise. Conversely, this technological progression is critically counterbalanced by an analysis of the systemic “side effects” of generative tools, addressing both technical-ethical dilemmas and critical cognitive risks, such as the potential erosion of critical thinking, autonomous problem-solving, empathy, and meaningful learning. Ultimately, this retrospective outlines a comprehensive conceptual roadmap for doctoral researchers, arguing that mitigating the risks of AIED requires transitioning from passive tool utility toward an integrated paradigm rooted in deep conceptual knowledge and critical awareness, thereby aligning algorithmic affordances with human-centred pedagogical goals. 

Short Bio: Sergio Miranda is a Senior Researcher in Experimental Pedagogy at the Department of Human Sciences, Philosophy and Education (DISUFF), University of Salerno, Italy. Characterized by a strong interdisciplinary academic background, he holds a degree in Computer Science and a Ph.D. in Information Engineering. In November 2025, he achieved the Italian National Scientific Habilitation (ASN) as a Full Professor in Didactics, Special Pedagogy, and Educational Research. His primary research lines lie at the intersection of Artificial Intelligence in Education (AIED), Technology-Enhanced Learning (TEL), Learning Analytics, and Knowledge Management. He is actively engaged in advanced postgraduate training as an elected member of the Doctoral Board for the Ph.D. program in Educational and Social Research: Society and Teaching-Learning Studies at the University of Salerno. On an international scale, he has served as an Executive Board member of the IEEE Technical Committee on Learning Technology, where he coordinated the Special Interest Group (SIG) on “Semantic Web in Education”. Furthermore, he is an Editorial Board member for several international peer-reviewed journals, including Education Sciences, AI for Society, and Artificial Intelligence and Education. Over the years, he has successfully served as the principal investigator and scientific coordinator for numerous competitive research frameworks, including EU-funded Erasmus+ projects and national initiatives dedicated to AI-driven skill mapping and immersive learning platforms. He is the author of more than 160 publications on e-learning technologies, knowledge management and artificial intelligence.

Emanuele Bellini

Department of Humanities Studies, Roma Tre University , Roma, Italy

12:00-13:00

Hybrid threats in Digital Cultural Heritage 

Short Bio: Prof. Emanuele Bellini is an Associate Professor in Information Processing Systems at the University of Roma Tre and a Visiting Academic at the University of Cambridge, Department of Computer Science and Technology. His research spans Cyber Resilience, Human-Cyber-Physical Systems, Critical Infrastructure Protection, and Trust Computing. He is also the founder of the emerging field of Cyber Humanities, and his current research investigates the security and protection of Digital Cultural Assets, as well as the counteraction of Cyber Cognitive Operations targeting Cultural Heritage, as part of the broader challenge of safeguarding memory,  knowledge and meaning in the digital age. Prof. Bellini serves as Chair of the IEEE SMC Technical Committee on Cyber Humanities and Vice-Chair of the IEEE SMC Technical Committee on Homeland Security. He is also Founder and Co-Chair of the IEEE International Conference on Cyber Security and Resilience (IEEE-CSR) and the IEEE International Conference on Cyber Humanities (IEEE-CH). His work is based on a transdisciplinary approch fusing technical and humanistic perspectives oto address the challenges of resilience, trust and sustainability in complex socio-technical critical ecosystems.