AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Ambient Dataloops: Generative Models for Dataset Refinement

Adrian Rodriguez-Munoz, William Daspit, Adam Klivans, Antonio Torralba, Constantinos Costis Daskalakis, Giannis Daras

ICML 2026 · Image generation › Fast diffusion sampling · approach: Distributional control of diffusion · 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 one Ambient Dataloops restoration loop on top of Ambient OmniAmbient Omni (loop 0) · CIFAR-10 with 90% corrupted data (blur and JPEG)unconditional FID: 5.689 -> 4.947 (sigma_B=0.6); up to 17% FID reductionImproved
treat corrupted data with ambient noise-level training instead of filtering or using all data as cleanCIFAR-10 90% corrupted · CIFAR-10 blur sigma_B=0.6FID: 11.26 (no filtering), 8.79 (filtering) -> 5.689 (Ambient Omni)Improved
add a second restoration pass using the loop-1 model (more loops) (ablation)Ambient Dataloops · CIFAR-10 blur sigma_B=0.6conditional metricsMixed
run 3 loops, reducing dataset noise by factor 8 each time (ablation)Ambient Dataloops · CIFAR-10 blur sigma_B=0.6FIDImproved
denoise dataset too fast or too slow, or beyond the model's capacity (ablation)Ambient Dataloops · CIFAR-10 blurFIDHurt
partially restore DiffusionDB to sigma 1.0 (half of 2.0) and retrainAmbient Omni text-to-image model · COCO zero-shotFID-30K: 10.61 -> 10.06Improved
partially restore DiffusionDB and retrainAmbient Omni text-to-image model · COCO zero-shotClip-FD-30K: 9.40 -> 8.83Improved
partially restore DiffusionDB and retrainAmbient Omni text-to-image model · GenEval overallscore: 0.47 -> 0.47No clear effect
increase SIMS guidance strength to improve image qualitySIMS guidance baseline · COCO zero-shotFID: 13.03 -> 13.23 (omega 0.0 -> 0.5)Hurt
restore bottom 20% of ImageNet with generative modelImageNet dataset · ImageNet quality metricsAesthetic score: 4.03 -> 4.41Improved
finetune ImageNet generator on restored ImageNetDegeorge et al. 2025 model · ImageNet generationImageReward, HPSv2, PickScore, Aesthetic: ImageReward -0.37 -> -0.19; PickScore 20.07 -> 20.13; Aesthetic 4.85 -> 4.81Mixed
apply one loop to AlphaFold protein backbone datasetAmbient Proteins · de novo protein backbone designdiversity vs designability: +14.3% diversity, -0.2% designabilityImproved

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