
Sergey Levine
Associate Professor · Electrical Engineering and Computer Sciences · University of California, Berkeley
At Prof Rod’s reading desk
Questions to explore
- How does a learning objective balance reward and exploration?
- What evidence would show a policy works beyond its training conditions?
Research interests
Sergey Levine studies reinforcement learning, robot learning and decision making.
- Reinforcement learning
- Robot learning
- Decision making
Explore the work
Prof Rod’s reading notes
Start with a paper
ICML · 2018
Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
Prof Rod’s reading note: Identify the role of entropy in the learning objective and the controls needed to compare policies across runs.
Bring a question of your own
Turn reading into research
Name the reward, policy and evaluation unit for a small agent task. Explain what would count as generalization rather than repeating a training scenario.
Prof Rod’s evaluation lesson helps you ask what was sampled and what a comparison supports. This is our suggested preparation, not the researcher’s admission requirement.
Learn to compare model resultsKeep exploring
More at University of California, Berkeley
- Allon Wagner
Artificial intelligence
- Anant Sahai
Machine learning · Artificial intelligence
- Armando Fox
Human-computer interaction
- Benjamin Recht
Artificial intelligence · Robotics
Improve this page
Suggest a correction, share a paper to start with, or point readers to a group page or opening.