AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

DomainFusion: Generalizing To Unseen Domains with Latent Diffusion Models

Yuyang Huang, Yabo Chen, Yuchen Liu, xiaopeng zhang, Wenrui Dai, Hongkai Xiong, Qi Tian

ECCV 2024 · Image generation › Autoregressive and latent generation · approach: Latent diffusion with learned tokenizers · 7 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 Latent Distillation loss on top of ERM (ablation)DomainFusion · OfficeHome (RegNetY-16GF)average accuracy: 73.7 -> 78.3Improved
add online generated augmentation data (L_gen) on top of ERM (ablation)DomainFusion · OfficeHome (RegNetY-16GF)average accuracy: 73.7 -> 78.6Improved
add Latent Distillation on top of generated augmentation (ablation)DomainFusion · OfficeHome (RegNetY-16GF)average accuracy: 78.6 -> 83.4Improved
use DomainFusion instead of best prior ResNet-50 methodResNet-50 · DG average over 5 datasetsaccuracy: 65.9 -> 67.5Improved
use DomainFusion instead of MIRO on RegNetY-16GF SWAGRegNetY-16GF · PACSaccuracy: 97.4 -> 96.6Hurt
use online lightweight augmentation instead of offline generation (CDGA)DomainFusion · DomainNetgenerated images: 4 million -> 120kImproved
add Latent Distillation to DomainFusion (ablation)DomainFusion · OfficeHomeProxy A-distanceImproved
use contrast reduction as corruption instead of blur, noise or random masking (ablation)Latent Distillation · supplementaryaccuracyImproved

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