Research map / Label-efficient and robust learning / Semi-supervised learning
Regularizing Discriminative Capability of CGANs for Semi-Supervised Generative Learning
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 |
|---|---|---|---|
| Add random regional replacement regularization on D1 (ablation) | Triangle-GAN on CIFAR-10 (4k) · CIFAR-10 (4k) | FID: 37.21 -> 25.30 | Helped |
| Add random regional replacement regularization on D1 (ablation) | Triangle-GAN on CIFAR-100 (10k) · CIFAR-100 (10k) | IS: 4.29 -> 7.02 | Helped |
| Add random regional replacement regularization on D1 (ablation) | Triangle-GAN on CIFAR-100 (10k) · CIFAR-100 (10k) | FID: 72.39 -> 31.18 | Helped |
| Add R3 regularization to the class-conditional GAN | R3-CGAN versus Baseline · SVHN (1k) | FID: 45.03 -> 10.87 | Helped |
| Add R3 regularization to the class-conditional GAN | R3-CGAN versus Baseline · CIFAR-10 (4k) | FID: 37.21 -> 20.34 | Helped |
| Add R3 regularization to the class-conditional GAN | R3-CGAN versus Baseline · CIFAR-100 (10k) | IS: 4.29 -> 7.49 | Helped |
| Add R3 regularization to the class-conditional GAN | R3-CGAN versus Baseline · FaceScrub-100 (2k) | FID: 31.21 -> 25.28 | Helped |
| Add R3 regularization to the class-conditional GAN | R3-CGAN versus Baseline · CIFAR-10 (4k) | test error rate (%): 13.51±0.58 -> 6.69±0.28 | Helped |
| Add R3 regularization to the class-conditional GAN | R3-CGAN versus Baseline · FaceScrub-100 (2k) | test error rate (%): 24.03±0.55 -> 6.96±0.43 | Helped |
| Remove R3 regularization on D1 (ablation) | R3-CGAN · CIFAR-10 (4k) | FID: 20.34 -> 32.68 | Hurt |
| Remove R3 regularization on D1 (ablation) | R3-CGAN · CIFAR-100 (10k) | FID: 26.29 -> 55.26 | Hurt |
| Remove R3 regularization on the classifer C (ablation) | R3-CGAN · CIFAR-10 (4k) | FID: 20.34 -> 22.52 | Hurt |
Closest papers by meaning
- Enhancing TripleGAN for Semi-Supervised Conditional Instance Synthesis and Classification (CVPR 2019 · similarity 0.91)
- Mask-Embedded Discriminator With Region-Based Semantic Regularization for Semi-Supervised Class-Conditional Image Synthesis (CVPR 2021 · similarity 0.91)
- Consistency Regularization for Generative Adversarial Networks (ICLR 2020 · similarity 0.90)
- Diversity-Sensitive Conditional Generative Adversarial Networks (ICLR 2019 · similarity 0.90)
- Triple Generative Adversarial Nets (NeurIPS 2017 · similarity 0.90)
- MS$^3$D: A RG Flow-Based Regularization for GAN Training with Limited Data (ICML 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.