Research map / Label-efficient and robust learning / Recognition and long-tail
Dual Adversarial Network for Deep Active Learning
Shuo Wang, Yuexiang Li, Kai Ma, Ruhui Ma, Haibing Guan, Yefeng Zheng
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add the representativeness discriminator (rDis) to the uncertainty-driven selector (full DAAL vs uDis-only) (ablation) | DAAL · CIFAR10/100 and Cityscapes | accuracy / mIoU | Improved |
| add the uncertainty discriminator (uDis) to the representativeness-only selector (ablation) | DAAL · CIFAR10/100 and Cityscapes | accuracy / mIoU | Improved |
| add the representativeness discriminator to Learning Loss without random subset strategy (ablation) | Learning Loss · CIFAR10/100 and Cityscapes | accuracy / mIoU | Improved |
| use a random subset strategy with Learning Loss (ablation) | Learning Loss · CIFAR10/100 and Cityscapes | accuracy / mIoU | Improved |
| combine Core-set with Learning Loss (multi-stage hybrid) (ablation) | Learning Loss · CIFAR10/100 and Cityscapes | accuracy / mIoU | Improved |
| feed task-model features to VAAL (VAAL + TF) (ablation) | VAAL · CIFAR10 | accuracy | Improved |
| add random subset selection to VAAL and VAAL + TF (ablation) | VAAL · CIFAR10/100 and Cityscapes | accuracy / mIoU | Improved |
| increase annotation budget size per iteration (ablation) | Learning Loss · CIFAR10/100 and Cityscapes | accuracy / mIoU | Improved |
| use DAAL instead of Learning Loss on CIFAR100 at 40% labeled data | active learning selector · CIFAR100 | accuracy (%): 39.15 -> 41.92 | Improved |
Closest papers by meaning
- Multi-Classifier Adversarial Optimization for Active Learning (AAAI 2023 · similarity 0.86)
- Deep Active Learning with Noise Stability (AAAI 2024 · similarity 0.85)
- Variational Adversarial Active Learning (ICCV 2019 · similarity 0.85)
- Provably Neural Active Learning Succeeds via Prioritizing Perplexing Samples (ICML 2024 · similarity 0.85)
- Revisiting Adversarial Training at Scale (CVPR 2024 · similarity 0.83)
- AutoAL: Automated Active Learning with Differentiable Query Strategy Search (ICML 2025 · similarity 0.83)
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.
