Luigi Acerbi

Luigi Acerbi is an Associate Professor in the Department of Computer Science at the University of Helsinki, where he leads the Machine and Human Intelligence research group. He is also an active member of the Finnish Center for Artificial Intelligence (FCAI) and ELLIS (European Laboratory for Learning and Intelligent Systems).

His research focuses on probabilistic machine learning and computational neuroscience, particularly on developing efficient methods for statistical inference, Bayesian models of perception, and resource-constrained rationality. His work bridges machine learning and cognitive science, with applications in Bayesian optimization, simulation-based inference, and image completion.

The recent publications highlight a strong trend toward unifying probabilistic conditioning across diverse tasks using transformer-based meta-learning frameworks like the Amortized Conditioning Engine (ACE). These works emphasize amortized inference, flexible latent variable modeling, and the integration of prior knowledge at runtime, enabling efficient and scalable Bayesian methods for complex problems.

• Scientific Affiliations:
  • University of Helsinki, Department of Computer Science
  • Finnish Center for Artificial Intelligence (FCAI)
  • ELLIS (European Laboratory for Learning and Intelligent Systems)
• Education:
  • PhD in Computational Neuroscience, Doctoral Training Centre, Edinburgh, UK
  • Advisor: Sethu Vijayakumar and Daniel Wolpert
  • Visiting work at Computational and Biological Learning Lab, Cambridge
• Postdoctoral Experience:
  • Alex Pouget’s lab, University of Geneva, Switzerland
  • Wei Ji Ma, New York University, USA
  • Collaboration with the International Brain Laboratory

Luigi Acerbi mentors PhD students including Daolang Huang and Nasrulloh Loka, and collaborates widely with researchers such as Samuel Kaski. He has contributed to open-source tools like PyVBMC and is involved in community initiatives such as the EurIPS conference. His work is supported by grants from the Research Council of Finland, Business Finland, and the UKRI Turing AI World-Leading Researcher Fellowship.

He leads a research lab focused on amortized probabilistic inference, with ongoing projects including PriorGuide and Stacked VBMC, aiming to make Bayesian methods more practical and accessible for real-world scientific and engineering applications.