AI Paper Atlas

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

Robust Reinforcement Learning with Bayesian Optimisation and Quadrature

Supratik Paul, Konstantinos Chatzilygeroudis, Kamil Ciosek, Jean-Baptiste Mouret, Michael A. Osborne, Shimon Whiteson

JMLR 2020 · Reinforcement learning › Policy gradient and actor-critic · approach: Model-based RL with guarantees · 1 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
use active Bayesian quadrature selection of theta instead of random sampling (ablation)ALOQ · Robot arm approximate p(theta)average cost: 25.95 -> 19.82Improved
use ALOQ policy over posterior instead of MAP point estimate of thetaALOQ · Robot arm approximate p(theta)average cost: 28.76 -> 19.82Improved
use ALOQ policy instead of RQ-ALOQ policy (ablation)ALOQ · Robot arm approximate p(theta)% episodes with cost >=70: 5.5% -> 0.2%Improved
add intensification (second evaluation of best policy) to ALOQ (ablation)ALOQ · Branin and Hartmann 6expected value of best policy: slightly underperforms one-step ALOQNo clear effect
add intensification to ALOQ (ablation)ALOQ · F-SRE1 and F-SRE2expected value of best policy: one-step ALOQ comparableNo clear effect
use ALOQ rather than WSN baselineALOQ · Braninexpected value of best policy: WSN substantially outperformedImproved
use ALOQ rather than WSN baselineALOQ · Hartmann 6expected value of best policy: WSN better than ALOQ variants except one-stepMixed
use ALOQ rather than naive and WSNALOQ · F-SRE1 and F-SRE2expected value of best policy: naive and WSN fail completelyImproved
use ALOQ rather than REINFORCE and TRPOALOQ · Robot arm collision avoidance and joint breakageexpected cost: converge to suboptimal policy even at 10000 simulator callsImproved
add active selection of simulator vs physical trials (TALOQ) over ALOQ trained only in simulationTALOQ · Physical robot ball strikereturn: 4.23 -> 93.60Improved
use TALOQ rather than ALOQ trained only on the proxy for realityTALOQ · Simulated robot arm (proxy for reality)sample efficiencyImproved
vary threshold k in TALOQ (ablation)TALOQ · Simulated robot armlearning speedNo 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