Logistics

• When: TBD (weekly, ~1 hour)
• Where: TBD
• Who: Open to all Cornell students and researchers, no formal RL background required, but familiarity with machine learning basics is helpful.

Format

1. The presenter gives a ~25–30 minute walkthrough of the paper: motivation, method, key results.
2. Open discussion follows: strengths, weaknesses, connections to other work, and open questions.
3. Slides are posted on the schedule page after each meeting.

Topics

Themes we expect to cover over the semester (subject to group interest):

• Foundations: value-based methods, policy gradients, actor-critic
• Exploration and sample efficiency
• Offline and off-policy RL
• RL from human feedback and preference-based learning
• Interactive learning, bandits, and imitation learning
• RL for and with large language models