Research map / Image generation / GANs
Active Generative Adversarial Network for Image Classification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 uncertainty-reward-guided conditional generation | SVM trained with labeled data plus generated images · CIFAR-10, MNIST, Fashion-MNIST, Tiny-ImageNet | F-score: 0.9 points and 1.0 points; 1.4 points and 1.1 points | Helped |
| add ordinary AC-GAN generated images to the labeled training set | SVM · CIFAR-10, MNIST, Fashion-MNIST, Tiny-ImageNet | F-score: 2.0 points and 2.5 points | Hurt |
| filter AC-GAN samples by smallest margin | AC-GAN+F · CIFAR-10, MNIST, Fashion-MNIST, Tiny-ImageNet | F-score | Mixed |
| add both smallest-margin and label-entropy rewards (ablation) | ActiveGAN · CIFAR-10 (n=10k) | F-score: 0.5 and 0.8 points | Helped |
| use label entropy alone instead of combining both uncertainty criteria (ablation) | ActiveGAN · CIFAR-10 (n=10k) | F-score: about 2.6 points compared with AC-GAN+F | Helped |
| use smallest-margin reward alone instead of combining both uncertainty criteria (ablation) | ActiveGAN · CIFAR-10 (n=10k) | F-score: 2.3 points | Helped |
| set ε to 0.2 rather than ε=1.0 (ablation) | ActiveGAN · CIFAR-10 (n=10k) | F-score: 86.3 vs 85.0 | Helped |
| remove reward truncation by setting ε to 1.0 (ablation) | ActiveGAN · CIFAR-10 (n=10k) | F-score: 86.3 -> 85.0 | Hurt |
Closest papers by meaning
- Activation Maximization Generative Adversarial Nets (ICLR 2018 · similarity 0.89)
- Multi-Classifier Adversarial Optimization for Active Learning (AAAI 2023 · similarity 0.89)
- Conditional GANs with Auxiliary Discriminative Classifier (ICML 2022 · similarity 0.88)
- A Unified Generative Adversarial Network Training via Self-Labeling and Self-Attention (ICML 2021 · similarity 0.88)
- Adversarial Data Programming: Using GANs to Relax the Bottleneck of Curated Labeled Data (CVPR 2018 · similarity 0.88)
- Fusing Conditional Submodular GAN and Programmatic Weak Supervision (AAAI 2024 · similarity 0.86)
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.