Welcome to the ALPI lab website!
The Autonomous Learning and Predictive Intelligence (ALPI, italian name for Alps :) ) lab works to build new algorithms to solve sequential decision-making problems. The research interest mostly focuses on theoretical Reinforcement Learning, Imitation Learning, and Multi-Agent learning, but we are also interested in every principled learning algorithm. With a commitment to creating reliable and efficient solutions, our ultimate goal is to address real-world challenges. Our works are presented at major machine-learning conferences such as NeurIPS, ICML, ICLR, AAMAS, AAAI, or AISTATS. See our latest publications here.
Our lab was founded by Giorgia Ramponi in 2024 and is part of the Department of Informatics at the University of Zurich.
Prospective Ph.D. students and postdocs can find application information here.
I’m very excited to share that I’ve been awarded an SNSF Starting Grant on multi-agent learning without explicit reward specification, leveraging human behavior and feedback, and embracing the idea that learning also emerges through interactions. Will advertise openings soon.
We are organizing the second edition of the Aligning Reinforcement Learning Experimentalists and Theorists workshop at NeurIPS 2025.
We will present four papers at NeurIPS, including one spotlight! See you in San Diego!
We had 8 papers presented at EWRL 2025! See you in Tübingen!
Giving two talks tomorrow at the Reinforcement Learning Conference in Edmonton on Imitation Learning and IRL in multi-agent systems.
One paper presented at ICLR 2025, see you in Singapore!
One paper presented at AAMAS 2025, see you in Detroit!
I am very happy to share that I got an Hassler Research Grant for the project: "Unified Feedback Integration Framework for Reinforcement Learning". With this grant we will start developing an unified framework to work with different kinds of feedback in RL, such as preferences, rewards, and demonstrations.