AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Post-training Iterative Hierarchical Data Augmentation for Deep Networks

Adil Khan, Khadija Fraz

NeurIPS 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Data augmentation and generalization bias · 10 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 IHDA fine-tuning on feature-space augmentationsResNet-110 · CIFAR-10 testerror (%): 6.45 -> 2.78Improved
add IHDA fine-tuningResNet-110 · CIFAR-100 testerror (%): 28.57 -> 18.45Improved
add standard input-space augmentation (IHDA+ vs IHDA) (ablation)IHDA ResNet-110 · CIFAR-10 testerror (%): 2.78 -> 2.72Improved
IHDA instead of AAA on Wide-ResNet-28-10Wide-ResNet-28-10 · CIFAR-10 testerror (%): 1.90 -> 2.19Hurt
fine-tune all subsequent layers instead of only the predictor (ablation)IHDA ResNet-110 · CIFAR-10 testerror (%): 6.01 -> 2.78Improved
augment at every level (p=0) versus level dropout p=0.55 (ablation)IHDA ResNet-110 · CIFAR-10 testerror (%): 6.87 -> 2.78Improved
decrease dropout probability for deeper levels instead of same p (ablation)IHDA ResNet-110 · CIFAR-10 testerror (%): 5.01 -> 2.78Improved
select source points by potential in hard regions instead of random (ablation)IHDA ResNet-110 · CIFAR-10 testerror (%): 5.73 -> 2.78Improved
select source points by potential instead of random (ablation)IHDA ResNet-110 · CIFAR-100 testerror (%): 26.91 -> 18.45Improved
IHDA versus training baseline for same extra epochs (ablation)ResNet-110 · CIFAR-10 testerror (%): 6.33 -> 2.78Improved
add IHDA to MLP classifierSHL network · SHL testaccuracy (%): 83% -> 92% ± 0.05%Improved
combine IHDA+ with AutoAugment policiesWide-ResNet-28-10 · CIFAR-10 testerror (%): 2.11 -> 1.92Improved

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