Scholar
Glen Berseth
Google Scholar ID: -WZcuuwAAAAJ
Assitant Professor - Université de Montréal
Citations & Impact
Citations
3,727
H-index
26
i10-index
46
Publications
20
Co-authors
31
list available
Publications
Resume (English only)
Academic Achievements
- 2025: Three papers accepted to ICLR 2025 on efficient imitation learning, exploration for GFNs, and real-time reinforcement learning
- 2024: Three papers accepted to NeurIPS 2024 on inverse safe RL, converging/scaling deep RL algorithms, and amortizing intractable inference
- 2024: Paper accepted to RSS 2024 on creating more diverse datasets for generalist robot policies
- 2024: Best paper award at ICLR 2024 on VLMs for robotics
- 2024: Five papers accepted to ICLR 2024 on efficient exploration, generalization in sequence planning, intelligent switching for reset-free RL, LLMs + RL for drug discovery, and latent diffusion for offline RL
- 2023: Paper accepted to NeurIPS 2023 on efficient exploration in finite horizons
- 2023: Paper accepted to iROS on using offline RL to bootstrap human-robot interfaces
- 2023: Paper accepted to RA-L on torque vs. position control for humanoids and Sim2Real training
- 2022: IROS paper on quadruped robots learning soccer shooting; IROS Best RoboCup Paper Award Finalist
- 2021: NeurIPS paper on surprise minimization in partially observed environments
- 2021: ICRA paper on RL for bipedal robots featured in MIT Technology Review
Background
- Assistant Professor at Université de Montréal
- Core academic member of Mila - Quebec AI Institute
- Canada CIFAR AI Chair
- Member of L'Institut Courtois
- Co-director of the Robotics and Embodied AI Lab (REAL)
- Research focuses on machine learning and real-world sequential decision-making problems (e.g., planning/RL) in robotics, scientific discovery, and adaptive clean technology
- Specific research areas include human-robot collaboration, generalization, reinforcement learning, continual learning, meta-learning, multi-agent learning, and hierarchical learning
- Teaches data science and robot learning courses at Université de Montréal and Mila
- Co-created a new conference for reinforcement learning research
- Supports the 'Slow Science' movement
