AI Paper Atlas

Research map / Learning theory and algorithms / Sample complexity

Learning to Guide Random Search

Ozan Sener, Vladlen Koltun

ICLR 2020 · Learning theory and algorithms › Sample complexity · approach: Sample complexity of learning distributions · 4 citations (OpenAlex)

Read the paper

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
learn the nonlinear search manifold online and combine manifold and full-space searchrandom search · MuJoCo locomotion suiteepisodes to threshold: LMRS uses fewer episodes than ARS on all six tasksImproved
learn the nonlinear search manifold online and combine manifold and full-space searchrandom search · Airfoil after 1500 XFoil simulationsLift-Drag: 0.560 -> 1.9737Improved
add online manifold learning during search (ablation)LMRS · MuJoCo tasks beyond Swimmerepisodes to threshold: Swimmer: 320 -> 222; Hopper: 5561 (4/5) -> 2408; HalfCheetah: 6528 -> 1128; Walker: 295493 (3/5) -> 15525; Ant: 24520 (4Improved
replace online manifold learning with offline manifold training (ablation)LMRS · MuJoCo tasksepisodes to threshold: Swimmer: 325 -> 222; Hopper: 11616 -> 2408; HalfCheetah: 3416 -> 1128; Walker: 118640 -> 15525; Ant: 63720 (2/5) -> 9240Mixed
remove random search and optimize using surrogate gradients alone (ablation)LMRS · MuJoCo locomotion taskstask-solving successHurt
reduce manifold dimensionality relative to ambient dimensionalityLMRS · synthetic optimization problemsobjective valueImproved
use LMRS rather than ARScontinuous optimization benchmarks · Pagmo suite, 46 functionsperformance profileImproved
use LMRS rather than REMBOcontinuous control · MuJoCo locomotion suitetask-solving successImproved

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