AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Learning to See by Looking at Noise

Manel Baradad, Jonas Wulff, Tongzhou Wang, Phillip Isola, Antonio Torralba

NeurIPS 2021 · Image generation › Autoregressive and latent generation · approach: Learning-based image quality assessment · 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
Pretrain with StyleGAN-Oriented synthetic noise rather than random initializationResNet-50 · ImageNet-1Ktop-1 accuracy: 4.36 -> 38.12Improved
Combine Dead leaves - Shapes, Feature Vis. - Dead leaves, Spectrum + Color + WMM, and StyleGAN-OrientedResNet-50 · ImageNet-1Ktop-1 accuracy: 38.12 -> 40.06Improved
Continuously sample StyleGAN-Oriented images instead of training on a fixed 1.3M imagesResNet-50 with StyleGAN-Oriented · ImageNet-1Ktop-1 accuracy: 38.12% -> 38.94%Improved
Add diverse synthetic procedural-noise data rather than use random CNN featuresResNet-50 · ImageNet-100top-1 accuracy: 10.84 -> 60.66Improved
Use higher-frequency noise maps and wavelet filters rather than default StyleGAN initialization (ablation)AlexNet · ImageNet-100top-1 accuracy: 0.34 -> 0.42Improved
Add sparse noise envelope and random convolution biases rather than default StyleGAN initialization (ablation)AlexNet · ImageNet-100top-1 accuracy: 0.34 -> 0.40Improved
Tie wavelet filters across output channels rather than use default StyleGAN initialization (ablation)AlexNet · ImageNet-100top-1 accuracy: 0.34 -> 0.43Improved
Add color statistics to spectrum-only statistical noise (ablation)AlexNet contrastive learner · ImageNet-100top-1 accuracy: 0.31 -> 0.37Improved
Add WMM to spectrum-and-color statistical noise (ablation)AlexNet contrastive learner · ImageNet-100top-1 accuracy: 0.37 -> 0.40Improved
Use oriented squares instead of axis-aligned squares in dead-leaves images (ablation)AlexNet contrastive learner · ImageNet-100top-1 accuracy: 0.35 -> 0.37Improved
Use varied shapes instead of axis-aligned squares in dead-leaves images (ablation)AlexNet contrastive learner · ImageNet-100top-1 accuracy: 0.35 -> 0.39Improved
Increase perceptual variation between crops beyond the reported sweet spot (ablation)Contrastive learning on synthetic datasets · ImageNet-100top-1 accuracyHurt

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