AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

OnlineAugment: Online Data Augmentation with Less Domain Knowledge

Zhiqiang Tang, Yunhe Gao, Leonid Karlinsky, Prasanna Sattigeri, Rogerio Feris, Dimitris Metaxas

ECCV 2020 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Geometric alignment and synthetic-to-real adaptation · 76 citations (OpenAlex)

Read the paper

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
use double instead of single cycle-consistency in noise-conditioned A-STN (ablation)A-STN · R-CIFAR-10accuracy (%): 82.62 -> 84.94Improved
use double instead of single cycle-consistency in image-conditioned A-STN (ablation)A-STN · R-CIFAR-10accuracy (%): 83.24 -> 84.76Improved
condition A-STN on Gaussian noise instead of image (double cycle) (ablation)A-STN · R-CIFAR-10accuracy (%): 84.76 -> 84.94No clear effect
add smoothness regularization to D-VAE (ablation)D-VAE · R-CIFAR-10accuracy (%): 82.72 -> 82.86No clear effect
add smoothness regularization to D-VAE (ablation)D-VAE · Liver segmentationDice (%): 68.51 -> 70.49Improved
use P-VAE noise generation instead of AdvProp with PGDP-VAE · R-CIFAR-10accuracy (%): 83.00 -> 84.08Improved
add A-STN on top of AutoAugmentAutoAugment (Wide-ResNet-28-10) · R-CIFAR-10accuracy (%): 85.43 -> 89.39Improved
combine A-STN, D-VAE and P-VAE on top of AutoAugmentAutoAugment (Wide-ResNet-28-10) · R-CIFAR-100accuracy (%): 47.87 -> 53.72Improved
combine all three models on Shake-Shake CIFAR-100 over best single modelAutoAugment (Shake-Shake) · R-CIFAR-100 (Shake-Shake)accuracy (%): 50.87 -> 50.11Hurt
add OnlineAugment on top of AutoAugmentAutoAugment · CIFAR-10top-1 error (%): 2.6 -> 2.0Improved
use OnlineAugment alone instead of AutoAugmentResNet-50 · ImageNettop-1 error (%): 22.4 -> 22.5No clear effect
combine A-STN and D-VAE instead of A-STN aloneOnlineAugment · LiTS (average Dice)Dice (%): 72.13 -> 73.35Improved

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