AI Paper Atlas

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

Fast Efficient Hyperparameter Tuning for Policy Gradient Methods

Supratik Paul, Vitaly Kurin, Shimon Whiteson

NeurIPS 2019 · Reinforcement learning › Policy gradient and actor-critic · approach: Trajectory reuse in policy gradients · 21 citations (OpenAlex)

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
Adapt the learning rate online with HOOFA2C · MuJoCo continuous control taskslearning speedImproved
Adapt the learning rate online with HOOFA2C · HalfCheetah and WalkerperformanceImproved
Use HOOF instead of meta-gradients with a grid-tuned meta learnerA2C · MuJoCo continuous control tasksperformance: 36 times as many samplesImproved
Vary the KL constraint epsilon across the tested range (ablation)HOOF with A2C · HalfCheetah, Hopper, Ant, and WalkerreturnNo clear effect
Remove the KL constraint (ablation)HOOF with A2C · MuJoCo continuous control tasksconvergenceHurt
Use a KL constraint to keep candidate policy updates boundedFirst-order HOOF · MuJoCo continuous control tasksWIS estimate reliabilityImproved
Use a (gamma, lambda)-conditioned value functionHOOF-TNPG · HalfCheetah, Hopper, and WalkerperformanceImproved
Remove (gamma, lambda) conditioning from the value function (ablation)HOOF-TNPG · HalfCheetah, Hopper, and WalkerperformanceHurt
Remove (gamma, lambda) conditioning from the value function (ablation)HOOF-TNPG · AntperformanceNo clear effect
Learn delta, gamma, and lambda with HOOFTNPG · HalfCheetah, Hopper, and Antlearning speedImproved
Learn delta, gamma, and lambda with HOOF instead of using default TRPO settingsTNPG · HalfCheetah, Hopper, and AntperformanceImproved
Learn delta, gamma, and lambda with HOOF instead of using default TRPO settingsTNPG · WalkerperformanceNo clear effect

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