During the week of 9–12 June 2025, the DaSCI Institute had the pleasure of welcoming Dr Angelo Porrello, a postdoctoral researcher at the University of Modena and Reggio Emilia (UNIMORE), who gave a specialised course entitled “Advanced Training Strategies for Incremental and Decentralised Learning”. This 10-hour training course offered an in-depth and practical overview of some of the most relevant current challenges in the field of machine learning.
Course content
The course was structured in three sessions spread throughout the week:
- Monday, 9 June
- Parameter-Efficient Fine-Tuning: techniques for adapting large models with minimal parameter updates.
- Continual Learning: strategies to mitigate catastrophic forgetting and enable continuous learning over time.
- Tuesday, 10 June
- Continuation of the Continual Learning block.
- Federated Learning and Model Merging: decentralised training and model aggregation techniques in distributed environments.
- Thursday, 12 June
- Mammoth Framework: introduction to this open-source library for evaluating continuous learning methods.
- Hands-on with Mammoth: practical session with code examples and use cases.
About Dr Angelo Porrello
Dr Angelo Porrello is part of the AImageLab research group in the Enzo Ferrari Department of Engineering at UNIMORE, where he has been conducting scientific work for eight years. He currently holds a postdoctoral position as a research assistant.
His research focuses on incremental and modular training paradigms for deep neural networks, as well as the development of predictive AI systems and video analysis, especially applied to video surveillance. He is the author of more than thirty peer-reviewed scientific articles, presented at top-level conferences and journals such as CVPR, ECCV, ICCV, NeurIPS, ICLR and T-PAMI. More than a dozen of these papers focus specifically on continual learning.
In addition to his research activity, Dr. Porrello is actively involved as a reviewer at international conferences and has co-edited a special issue for the MDPI journal on the use of AI in infectious disease monitoring. He is a member of ELLIS (European Laboratory for Learning and Intelligent Systems), a European network of excellence in machine learning.
Relevance to our community
Dr. Porrello’s visit provided an exceptional opportunity for specialised training for DaSCI researchers. The topics covered—such as continuous learning, federated training, and efficient model adaptation—are key areas for advancing toward more sustainable, scalable, and robust artificial intelligence systems. In addition, the practical approach of the course allowed attendees to familiarise themselves with open-source tools such as Mammoth, facilitating their incorporation into ongoing research projects.
DaSCI would like to thank Dr. Porrello for his generosity in sharing his knowledge and experience, and we look forward to continuing to collaborate in the future within the framework of networks such as ELLIS or joint projects.
This training is part of the project “Ethical, Responsible and General Purpose Artificial Intelligence: Applications in Risk Scenarios. (IAFER) Ref.: TSI-100927-2023-1, funded through the creation of university-business chairs (Enia Chairs) for research and development in artificial intelligence, for dissemination and training within the framework of the European Recovery, Transformation and Resilience Plan, funded by the European Union-Next Generation EU.
