Research map / Label-efficient and robust learning / Recognition and long-tail
EntAugment: Entropy-Driven Adaptive Data Augmentation Framework for Image Classification
Suorong Yang, Furao Shen, Jian Zhao
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add EntLoss to EntAugment (ablation) | EntAugment · CIFAR-100 ResNet-50 | accuracy (%): 81.56 -> 82.49 | Improved |
| add EntLoss to EntAugment (ablation) | EntAugment · CIFAR-10 ResNet-44 | accuracy (%): 95.76 -> 95.58 | Hurt |
| add EntLoss to Random Erasing (ablation) | Random Erasing · CIFAR-100 ResNet-50 | accuracy (%): 77.79 -> 79.56 | Improved |
| add EntLoss to Random Erasing (ablation) | Random Erasing · CIFAR-100 ResNet-18 | accuracy (%): 75.97 -> 75.98 | No clear effect |
| add EntLoss to FAA (ablation) | FAA · CIFAR-10 ResNet-50 | accuracy (%): 96.69 -> 96.58 | Hurt |
| add EntLoss to TrivialAugment (ablation) | TrivialAugment · CIFAR-100 ResNet-18 | accuracy (%): 78.67 -> 79.48 | Improved |
| reduce number of base transformations M from 14 to 2 (ablation) | EntAugment · CIFAR-10 ResNet-18 | accuracy: 96.84% -> 96.36% | Hurt |
| apply EntAugment and EntLoss to OLTR | OLTR on ImageNet-LT · ImageNet-LT closed-set | overall top-1 accuracy (%): 35.6 -> 39.6 | Improved |
| apply EntAugment versus MADAug on ImageNet | EntAugment · ImageNet ResNet-50 | top-1 accuracy (%): 78.5 -> 78.2 | Hurt |
| add EntLoss to training with EntAugment (feature clustering) (ablation) | ResNet-18 on CIFAR-10 · CIFAR-10 | Dunn index: 7.93e-5 -> 1.07e-4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TeachAugment — TeachAugment: Data Augmentation Optimization Using Teacher Knowledge (CVPR 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Data-Efficient Augmentation for Training Neural Networks (NeurIPS 2022 · similarity 0.87)
- Inversion Circle Interpolation: Diffusion-based Image Augmentation for Data-scarce Classification (CVPR 2025 · similarity 0.86)
- When Dynamic Data Selection Meets Data Augmentation: Achieving Enhanced Training Acceleration (ICML 2025 · similarity 0.86)
- Effective Data Augmentation With Diffusion Models (ICLR 2024 · similarity 0.85)
- Diversify, Contextualize, and Adapt: Efficient Entropy Modeling for Neural Image Codec (NeurIPS 2024 · similarity 0.85)
- DiffAug: A Diffuse-and-Denoise Augmentation for Training Robust Classifiers (NeurIPS 2024 · similarity 0.85)
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.
