معرفی
Stefano Melacci is an Associate Professor in the Department of Information Engineering and Mathematics (DIISM) at the University of Siena, Italy. He is a core member of the Siena Artificial Intelligence Laboratory (SAILab), where his research centers on Machine Learning and Artificial Intelligence, particularly in foundational and applied aspects of Computer Vision and Natural Language Processing.
- Department: Department of Information Engineering and Mathematics (DIISM)
- Institution: University of Siena
- Research Group: Siena Artificial Intelligence Laboratory (SAILab)
He holds a PhD in Machine Learning from the University of Siena (2010), with prior research experience at Ohio State University. His career includes academic positions and industrial research at QuestIT, where he led NLP and conversational AI initiatives.
His research interests lie at the intersection of deep learning and symbolic reasoning, particularly through the framework of Learning from Constraints. He investigates Kernel Machines, Regularization Theory, and cognitive models such as Developmental Visual Agents (DVA) and Cognitive Action Laws (CALs). His work emphasizes motion invariance, visual attention, and continual learning.
The recent publications highlight a strong trend in neurosymbolic AI, constraint-based learning, and biologically inspired neural computation. Key themes include explainability, adversarial robustness, graph neural networks, and efficient temporal modeling. His collaborative work with Marco Gori, Andrea Betti, and others demonstrates sustained innovation in hybrid AI systems.
Scientific Contributions and Recognition:
- Author of foundational work on Laplacian SVMs trained in the primal (JMLR 2011)
- Co-author of the book Machine Learning: A Constraint-Based Approach (Elsevier, 2nd ed. 2024)
- Developer of PyTorch and MATLAB libraries for LapSVMs and cognitive vision models
- Active contributor to top-tier venues: NeurIPS, ICML, AAAI, IJCAI, IEEE TPAMI
Stefano Melacci has advised numerous students in deep learning, continual learning, and neurosymbolic AI. His research is supported by sustained publication output and integration of theoretical and applied work. He teaches courses such as Neural Networks and Fondamenti di Programmazione at the University of Siena.
He leads research within SAILab, focusing on cognitive vision, constraint propagation, and hybrid learning systems. His lab develops models that combine data-driven learning with symbolic knowledge, aiming to create more robust, interpretable, and human-like AI systems.
حوزههای پژوهشی
