AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Improved Masked Image Generation with Knowledge-Augmented Token Representations

Guotao Liang, Baoquan Zhang, Zhiyuan Wen, Zihao Han, Yunming Ye

AAAI 2026 · Image generation › Autoregressive and latent generation · approach: Generative latent space tooling · 0 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
add knowledge-augmented token graph priors (KA)MaskGIT · ImageNet 256x256 class-conditionalFID: 6.18 -> 5.69Improved
add knowledge-augmented token graph priors (KA)MaskGIT · ImageNet 256x256 class-conditionalIS: 182.1 -> 170.2Hurt
add KA graph priorsAutoNAT · ImageNet 256x256 class-conditionalFID: 2.68 -> 2.45Improved
add KA graph priorsTiTok-s128 · ImageNet 256x256 class-conditionalFID: 1.97 -> 1.90Improved
add KA graph priors to shorter-token-sequence modelTiTok-b64 · ImageNet 256x256 class-conditionalFID: 2.48 -> 2.40Improved
add Gco and Gs graph priorsMaskGen · MS-COCO 30K text-to-imageFID: 22.27 -> 21.01Improved
add semantic similarity graph alone (ablation)AutoNAT · ImageNet 256x256 (graph ablation)FID: 2.68 -> 2.49Improved
add incompatibility graph alone (ablation)AutoNAT · ImageNet 256x256 (graph ablation)FID: 2.68 -> 2.51Improved
add co-occurrence graph alone (ablation)AutoNAT · ImageNet 256x256 (graph ablation)FID: 2.68 -> 2.51Improved
combine all three graphs vs best single graph (ablation)AutoNAT · ImageNet 256x256 (graph ablation)FID: 2.49 -> 2.45Improved
add all three graphs (ablation)AutoNAT · ImageNet 256x256 (graph ablation)IS: 278.8 -> 274.1Mixed
add incompatibility graph (ablation)AutoNAT · Token rankingNDCG@100: 0.0282 -> 0.0284 (mask 0.3), 0.0288 -> 0.0289 (mask 0.5)No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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