AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

Projections for Approximate Policy Iteration Algorithms

Riad Akrour, Joni Pajarinen, Jan Peters, Gerhard Neumann

ICML 2019 · Reinforcement learning › Policy gradient and actor-critic · approach: Model-based RL with guarantees

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add projection-based extra optimization to PPO (PAPI-PPO)PPO · RSHopper-v1initial 100 iteration reward: 68 ± 8 -> 121 ± 8Improved
add projection-based extra optimization to PPOPPO · RSHalfCheetah-v1initial 100 iteration reward: 30 ± 2 -> 55 ± 4Improved
add extra optimization steps and projection to TRPO (PAPI-TRPO)TRPO · RSHopper-v1initial reward: 87 ± 11 -> 75 ± 7Hurt
add extra L∘g optimization steps on top of entropy-constrained TRPO (PAPI-TRPO vs PAPI-0-TRPO) (ablation)PAPI-0-TRPO · RSHalfCheetah-v1initial reward: 20 ± 2 -> 32 ± 4Improved
entropy-constrained TRPO without extra optimization (PAPI-0-TRPO) vs TRPO (ablation)TRPO · RSWalker-v1best-window reward: 1183 ± 372 -> 1633 ± 342Improved
PAPI-PPO vs PPOPPO · RSWalker-v1best-window reward: 1016 ± 359 -> 1426 ± 434Improved
PAPI-TRPO vs TRPOTRPO · RSHalfCheetah-v1best-window reward: 1882 ± 172 -> 2524 ± 113Improved
PAPI-PPO vs PPOPPO · RSAnt-v1best-window reward: 1871 ± 86 -> 1924 ± 110No clear effect
optimize L∘g by gradient ascent instead of projected gradient with same projection (ablation)Projection g · linear problem with entropy constraint (100 runs)regret and KL to optimalImproved
direct projection-based policy search (ProjectionPS) vs REPS and MORE (ablation)ProjectionPS · smooth mixture-of-25-Gaussians functionsperformance at low sample countsImproved
hard entropy constraint with linearly decreasing beta vs entropy bonus (ablation)PPO / TRPO · BitFlippolicy returnImproved
more optimization steps of L∘g on the policy update objective (ablation)TRPO / PPO · RoboschoolHopper policy updateobjective improvement and KLImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy