Ronald Parr
Ronald Parr studies reinforcement learning and large sequential decision problems. He earned a Berkeley PhD in 1998, worked at Stanford and joined Duke in 2000. With Michail Lagoudakis, he published Least-Squares Policy Iteration in 2003. The method learns state-action values and reuses sample experience to improve policies. He also coauthored algorithms for factored Markov decision processes. AAAI elected him a fellow in 2023 for several reinforcement-learning methods.
- Now at
- Duke University
- Works
- 130
- Citations
- 5,778
- h-index
- 38
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
1998–2000
Learning and control research
Parr received his computer science PhD at Berkeley in 1998, advised by Stuart Russell. His thesis studied hierarchical control and learning in Markov decision processes. He then worked with Daphne Koller at Stanford and joined Duke as an assistant professor in 2000.
-
2003
Least-squares policy iteration
With Michail Lagoudakis, Parr introduced Least-Squares Policy Iteration. It learns a state-action value function and improves the policy through repeated iterations. The model-free, off-policy method can reuse sample experiences, separating data collection from the approximation and solution methods.
-
2003
Factored decision problems
He coauthored Efficient Solution Algorithms for Factored MDPs with Carlos Guestrin, Daphne Koller and Shobha Venkataraman. The paper later received the IJCAI-JAIR award. It forms another part of his work on large stochastic planning problems.
-
2023
Recognition in reinforcement learning
AAAI elected Parr a fellow in 2023. The society cites foundational hierarchical, least-squares, factored and feature-selection methods. His research interests also connect sequential decisions with game theory, sensing and robotics.
Sources and credits
Photo: users.cs.duke.edu; Ronald Parr profile photograph; creator not stated. License: Not stated by source. Source: users.cs.duke.edu. Cropped to a square and resized. Ask us to remove or replace this photo.