Scholar
Mirco Mutti
Google Scholar ID: GlLkJ9UAAAAJ
Technion
Citations & Impact
Citations
409
H-index
10
i10-index
10
Publications
20
Co-authors
29
list available
Publications
Resume (English only)
Academic Achievements
- Paper 'Blindfolded experts generalize better' awarded Best Paper at EXAIT workshop, ICML 2025
- Paper 'A theoretical framework for partially-observed reward states in RLHF' accepted at ICLR 2025
- Preprint 'Reward compatibility: A framework for inverse RL' explores theoretical foundations of inverse RL
- Paper 'How does inverse RL scale to large state spaces? A provably efficient approach' accepted at NeurIPS 2024
- Paper 'The limits of pure exploration in POMDPs: When the observation entropy is enough' accepted at RLC conference
- Four papers accepted at ICML 2024 on meta RL, inverse RL, geometric active exploration, and pure exploration in POMDPs
- Paper 'A tale of sampling and estimation in discounted reinforcement learning' accepted with oral presentation at AISTATS 2023
- Paper 'The importance of non-Markovianity in maximum state entropy exploration' received Outstanding Paper Award at ICML 2022
- Served as co-program chair for EWRL 2022 in Milan
- Invited as a 'Rising star in AI' speaker at KAUST
Research Experience
- Postdoctoral researcher at Technion - Israel Institute of Technology since September 2023
- Working in the Robot Learning Lab with Aviv Tamar
- Research theme: 'Reinforcement learning from theory to practice'
- Gave a talk on 'Unsupervised reinforcement learning' at VANDAL lab in Turin
- Presented on '(Non)convex reinforcement learning' at ETH Zurich's LAS group and AI Center
- Toured northern Italy (MaLGa, IIT Genova, Bocconi, University of Verona) to present recent RL work
Education
- PhD from Politecnico di Milano
- Advised by Marcello Restelli
- Member of the Artificial Intelligence and Robotics Lab
- Successfully defended PhD thesis in March 2023
- PhD thesis received an honorable mention for best AI thesis by AIxIA (Italian Association for Artificial Intelligence)
Background
- Research interests center on reinforcement learning (RL)
- Current research focuses on generalization and meta RL
- Previous work emphasized unsupervised RL and learning without rewards
- Aims to advance theoretical understanding to enable real-world applications of RL
- Research topics include partial observability, RL with general utilities, RLHF, imitation learning, and inverse RL
