Research map / Label-efficient and robust learning / Semi-supervised learning
OSSGAN: Open-Set Semi-Supervised Image Generation
Check 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 |
|---|---|---|---|
| Use continuous classifier probability vectors as unlabeled-sample conditions, with entropy regularization | OSSGAN · Tiny ImageNet, limited-label configurations | FID: 60.09 -> 35.39 | Helped |
| Use continuous classifier probability vectors as unlabeled-sample conditions, with entropy regularization | OSSGAN · ImageNet | FID: 117.78 -> 78.43 | Helped |
| Add entropy regularization (ablation) | OSSGAN · Tiny ImageNet | FID: 70.55 -> 35.39 | Helped |
| Add entropy regularization (ablation) | OSSGAN · ImageNet | FID: 85.65 -> 78.43 | Helped |
| Add fake-sample cross-entropy (ablation) | OSSGAN · Tiny ImageNet | FID: 60.06 -> 35.39 | Helped |
| Add fake-sample cross-entropy (ablation) | OSSGAN · ImageNet | FID: 81.53 -> 78.43 | Helped |
| Add entropy regularization and fake-sample cross-entropy together (ablation) | OSSGAN · Tiny ImageNet | FID: 60.09 -> 35.39 | Helped |
| Assign low-confidence unlabeled samples to an added open-set class using a threshold | OpensetGAN · Tiny ImageNet | FID | Hurt |
| Discard low-confidence unlabeled samples using a threshold | RejectGAN · Tiny ImageNet | FID | Hurt |
| Assign high-entropy labels to open-set samples | Entropy oracle method · Tiny ImageNet | FID | Helped |
| Use fake-sample cross-entropy without entropy regularization (ablation) | OSSGAN · Tiny ImageNet with sufficient labeled samples | FID | Hurt |
Closest papers by meaning
- CcGAN: Continuous Conditional Generative Adversarial Networks for Image Generation (ICLR 2021 · similarity 0.91)
- High-Fidelity Image Generation With Fewer Labels (ICML 2019 · similarity 0.90)
- Regularizing Discriminative Capability of CGANs for Semi-Supervised Generative Learning (CVPR 2020 · similarity 0.89)
- A Unified View of cGANs with and without Classifiers (NeurIPS 2021 · similarity 0.89)
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 2024 · similarity 0.88)
- Fusing Conditional Submodular GAN and Programmatic Weak Supervision (AAAI 2024 · similarity 0.88)
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.