Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Trust Region Policy Optimization

About

We describe an iterative procedure for optimizing policies, with guaranteed monotonic improvement. By making several approximations to the theoretically-justified procedure, we develop a practical algorithm, called Trust Region Policy Optimization (TRPO). This algorithm is similar to natural policy gradient methods and is effective for optimizing large nonlinear policies such as neural networks. Our experiments demonstrate its robust performance on a wide variety of tasks: learning simulated robotic swimming, hopping, and walking gaits; and playing Atari games using images of the screen as input. Despite its approximations that deviate from the theory, TRPO tends to give monotonic improvement, with little tuning of hyperparameters.

John Schulman, Sergey Levine, Philipp Moritz, Michael I. Jordan, Pieter Abbeel• 2015

Related benchmarks

TaskDatasetResultRank
Reinforcement LearningWalker
Average Returns229.8
38
QuadrupedQuadruped
Return206.2
33
Reinforcement LearningHumanoid
Zero-Shot Reward3.18e+3
30
Reinforcement LearningTrading
Return9.17
24
Reinforcement Learningcheetah
Return210.1
24
Reinforcement LearningLunarLander v2
Final Return104
23
Reinforcement LearningHalfcheetah
Average Return1.77e+3
17
Robot LocomotionHumanoid
Cumulative Reward1.39
16
Reinforcement LearningPendulum
Avg Episode Reward-145.5
15
Reinforcement LearningHopper
Avg Episode Reward2.59e+3
15
Showing 10 of 69 rows

Other info

Follow for update