AI Paper Atlas

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

Batch Reinforcement Learning With a Nonparametric Off-Policy Policy Gradient

Samuele Tosatto, Joao Carvalho, Jan Peters

TPAMI 2021 · Reinforcement learning › Policy gradient and actor-critic · approach: Variance-reduced policy gradients

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
use NOPG nonparametric gradient instead of importance-sampling G(PO)MDP on off-policy LQG datagradient estimator · LQG gradient analysisbias and variance of gradient estimate: significantly both lower bias and varianceImproved
increase off-policiness alpha of the dataset (ablation)importance-sampling G(PO)MDP · LQG gradient analysisvarianceHurt
increase off-policiness alpha of the dataset (ablation)NOPG · LQG gradient analysisbiasHurt
decrease kernel bandwidth (ablation)NOPG-S · LQG gradient analysis (1000 samples, alpha 0.5)varianceHurt
increase kernel bandwidth (ablation)NOPG-S · LQG gradient analysis (1000 samples, alpha 0.5)biasHurt
policy moves toward low-density regions (nonparametric gradient)NOPG · Pendulum-v0 trust region analysisgradient magnitudeHurt
use NOPG instead of DDPG, TD3, SACpolicy optimization · Pendulum-v0 and Cart-Polesample efficiency: more than one order of magnitudeImproved
use NOPG instead of BEAR, BRAC, MOPO, MOReLoffline policy optimization · Pendulum-v0, Cart-Pole, U-Mazesample efficiencyMixed
use NOPG instead of behavioral cloning on human demonstrationsoffline learning · Mountain Car (human demonstrations)return: optimal policy with two demonstrated trajectoriesImproved
apply NOPG to higher-dimensional taskNOPG · Hopper (D4RL)returnNo 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