Research map / Label-efficient and robust learning / Semi-supervised learning
Good Semi-supervised Learning That Requires a Bad GAN
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 |
|---|---|---|---|
| replace FM with the proposed complement-generator method | Semi-supervised classifier · MNIST | # errors: 93 ± 6.5 -> 79.5 ± 9.8 | Helped |
| replace FM with the proposed complement-generator method | Semi-supervised classifier · SVHN | % errors: 8.11 ± 1.3 -> 4.25 ± 0.03 | Helped |
| replace FM with the proposed complement-generator method | Semi-supervised classifier · CIFAR-10 | % errors: 18.63 ± 2.32 -> 14.41 ± 0.30 | Helped |
| add input-space entropy maximization (VI) to FM (ablation) | FM · MNIST | Error: 85.0 ± 11.7 -> 86.5 ± 10.6 | Hurt |
| add low-density enforcement (LD) to FM (ablation) | FM · MNIST | Error: 85.0 ± 11.7 -> 79.5 ± 9.8 | Helped |
| add conditional entropy (Ent) to FM+LD (ablation) | FM+LD · MNIST | Error: 79.5 ± 9.8 -> 89.2 ± 10.5 | Hurt |
| add input-space entropy maximization (VI) to FM (ablation) | FM · CIFAR-10 | Error: 16.14 -> 14.41 | Helped |
| add conditional entropy (Ent) to FM+VI (ablation) | FM+VI · CIFAR-10 | Error: 14.41 -> 15.82 | Hurt |
| add input-space entropy maximization (VI) to FM (ablation) | FM · SVHN | Error: 6.83 -> 5.29 | Helped |
| add feature-space pull-away term (PT) to FM (ablation) | FM · SVHN | Error: 6.83 -> 4.63 | Helped |
| add conditional entropy (Ent) to FM+PT (ablation) | FM+PT · SVHN | Error: 4.63 -> 4.25 | Helped |
| add low-density enforcement (LD) to FM+PT+Ent (ablation) | FM+PT+Ent · SVHN | Error: 4.25 -> 4.19 | Helped |
Closest papers by meaning
- MarginGAN: Adversarial Training in Semi-Supervised Learning (NeurIPS 2019 · similarity 0.92)
- Semi-supervised Learning with GANs: Manifold Invariance with Improved Inference (NeurIPS 2017 · similarity 0.90)
- Enhancing TripleGAN for Semi-Supervised Conditional Instance Synthesis and Classification (CVPR 2019 · similarity 0.89)
- Triple Generative Adversarial Nets (NeurIPS 2017 · similarity 0.89)
- Teaching a GAN What Not to Learn (NeurIPS 2020 · similarity 0.88)
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 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.