AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Learning to Paint With Model-Based Deep Reinforcement Learning

Zhewei Huang, Wen Heng, Shuchang Zhou

ICCV 2019 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Imitation and skill learning · 186 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 model-based DDPG with a differentiable neural renderer instead of model-free DDPG with PatchQ (ablation)Learning to Paint agent · CelebA paintingtest l2 distance: 5x smaller l2 distanceImproved
use model-based DDPG instead of original DDPG (ablation)Learning to Paint agent · CelebA paintingtest l2 distance: 20x smaller l2 distanceImproved
add PatchQ patch-level rewards to model-free DDPG critic (ablation)DDPG · CelebA paintingtest performanceImproved
use WGAN discriminator reward instead of l2 reward (ablation)Learning to Paint agent · CelebA paintingtest l2 distanceImproved
increase number of strokes (100, 200, 400, 1000) (ablation)Learning to Paint agent · CelebA paintingtest lossImproved
use action bundle of 5 strokes per step (ablation)Learning to Paint agent · CelebA paintingtest lossImproved
compare with SPIRAL using 20 opaque strokesSPIRAL baseline · CelebA paintingtest l2 distance: 3x smallerImproved
replace quadratic Bezier strokes with straight strokes, circles or triangles (ablation)Learning to Paint agent · CelebA paintingpainting qualityNo 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