AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Learning AND–OR Templates for Compositional Representation in Art and Design

Liaoruxing Zhang, Xin Jin, Wenbo Yuan, ChenYu Fan, Song-Chun Zhu

ICLR 2026 · Image generation › Autoregressive and latent generation · approach: Compositional scene generation

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
add late fusion of template configuration features to VGG19VGG19 · AVA accuracyaccuracy (%): 78.08 -> 81.43Improved
add late fusion of template configuration features to ResNet50ResNet50 · AVA accuracyaccuracy (%): 85.71 -> 85.87Improved
add late fusion of template configuration features to ViT-B/16ViT-B/16 · AVA accuracyaccuracy (%): 85.38 -> 86.20Improved
add late fusion of template features to VGG19VGG19 · AVASRCC: 0.7506 -> 0.7893Improved
add late fusion of template features to ResNet50ResNet50 · AVASRCC: 0.7842 -> 0.8156Improved
add late fusion of template features to ViT-B/16ViT-B/16 · AVASRCC: 0.7590 -> 0.7855Improved
use template consistency score instead of a deep 2-class classifier (ResNet50) for correlation with human ratingsAND-OR templates · AVASRCC: 0.7842 -> 0.8419Improved
use templates instead of ResNet50 for AVA accuracyAND-OR templates · AVAaccuracy (%): 85.71 -> 85.65No clear effect
use templates instead of CNN baselines on a 10-class Places365 subsetAND-OR templates · Places365 (10-class subset)Top-5 error (%): 8.61 -> 8.03Improved
perturb templates by part substitution, relation shuffling or geometric jitter (ablation)consistency score · CITD themesconsistency scoreHurt
increase training images in semi-supervised learning (ablation)template AND-OR graph · AVA/AADB semi-supervisednumber of nodesMixed
increase training sample size n to 100 in semi-supervised learning (ablation)semi-supervised templates · cross-validation KLKL divergence: reduced to 0.1Improved

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