Antonio Vergari

Antonio Vergari serves as a Reader (equivalent to Associate Professor) in the School of Informatics at the University of Edinburgh, affiliated with the Institute for Adaptive and Neural Computation (ANC). His research develops probabilistic machine learning systems that are provably reliable in real-world applications through the integration of complex reasoning, efficient inference, and neuro-symbolic learning paradigms.

His primary research domains include Artificial Intelligence, Machine Learning, Probabilistic Modeling, and Neuro-symbolic AI, with specific expertise in probabilistic circuits, tensor networks, and constraint-satisfying architectures. Current investigations focus on unifying tensor factorization theory with circuit representations to overcome expressiveness limitations while maintaining computational tractability for complex reasoning tasks.

Analysis of his publication trajectory reveals consistent advancement in reliable probabilistic modeling, particularly through circuit-based approaches that guarantee logical consistency in neural predictions. His work bridges theoretical foundations (e.g., expressiveness hierarchies) with practical implementations (e.g., Cirkit library), addressing critical gaps in benchmark complexity and scalable constraint satisfaction.

Scientific recognition includes:

• ICLR 2024 Spotlight presentation (top 5% acceptance rate)
• NeurIPS 2023 Oral presentation (top 0.6% acceptance rate)
• NeurIPS 2021 Oral presentation (top 0.6% acceptance rate)

Vergari leads the APRIL Lab, which actively contributes to open-source probabilistic modeling tools like Cirkit while organizing community initiatives such as the CoLoRAI workshop at AAAI-25. The lab maintains strong industry and academic collaborations focused on developing theoretically-grounded, deployable probabilistic systems.