Daniele Berardini is a Postdoctoral Researcher at the Italian Institute of Technology (IIT), within the AIGO – AI for Good research line led by Prof. Vittorio Murino. He received his Ph.D. in Information Engineering (cum laude) from Università Politecnica delle Marche, where he conducted research on efficient deep learning methods for real-time human behavior analysis and visual perception systems, with a focus on lightweight and edge AI solutions.
His current research focuses on the development of principled machine learning methodologies for learning from heterogeneous, multimodal, and imperfect data. His work spans federated and distributed learning, representation alignment via Optimal Transport, domain adaptation and generalization, multi-task and cross-domain learning, and knowledge distillation under limited and no-data regimes, with the goal of enabling robust and scalable learning under strong data heterogeneity, limited supervision, and privacy constraints.
His work combines methodological advances with interdisciplinary applications, including multimodal human behavior and social signal analysis, biomedical and clinical imaging for data-driven precision medicine, privacy-preserving learning in distributed settings, and safety-critical perception systems. He has been involved in international research projects as a Marie Skłodowska-Curie visiting fellow and actively supervises graduate students and serves as reviewer for international journals and conferences.