Research map / Image generation / GANs
Triple Generative Adversarial Nets
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 |
|---|---|---|---|
| Use Triple-GAN rather than the strongest reported baseline | semi-supervised classification · MNIST, n=100 | error rate (%): 0.89 ($\pm 0.50$) vs 0.91 ($\pm 0.58$) | No clear effect |
| Use Triple-GAN rather than MMCVA | semi-supervised classification · SVHN, n=1000 | error rate (%): 4.95 ($\pm0.18$) $^\dagger$ vs 5.77($\pm0.17$) | Hurt |
| Use Triple-GAN rather than Improved-GAN | semi-supervised classification · SVHN, n=1000 | error rate (%): 8.11 ($\pm1.3$) -> 5.77($\pm0.17$) | Helped |
| Use Triple-GAN rather than Improved-GAN | semi-supervised classification · CIFAR10, n=4000 | error rate (%): 18.63 ($\pm2.32$) -> 16.99 ($\pm 0.36$) | Helped |
| Use Triple-GAN rather than Improved-GAN | semi-supervised classification · MNIST, n=20 | error rate (%): 16.77 ($\pm 4.52$) -> 4.81 ($\pm 4.95$) | Helped |
| Use Triple-GAN rather than Improved-GAN | semi-supervised classification · MNIST, n=50 | error rate (%): 2.21 ($\pm 1.36$) -> 1.56 ($\pm 0.72$) | Helped |
| Use Triple-GAN rather than Improved-GAN | semi-supervised classification · MNIST, n=200 | error rate (%): 0.90 ($\pm 0.04$) -> 0.67 ($\pm 0.16$) | Helped |
| Use Triple-GAN rather than Improved-GAN without minibatch discrimination | image generation · CIFAR10 | inception score: $3.87 \pm 0.03$ -> $5.08 \pm 0.09$ | Helped |
| Add the pseudo discriminative loss (ablation) | Triple-GAN · SVHN, n=1000 | error rate (%): about $7.8\%$ -> 5.77($\pm0.17$) | Helped |
| Add the generator and discriminator to a classifier-only baseline (ablation) | SVHN semi-supervised classification · SVHN | error rate (%) | Helped |
Closest papers by meaning
- MCL-GAN: Generative Adversarial Networks with Multiple Specialized Discriminators (NeurIPS 2022 · similarity 0.91)
- Enhancing TripleGAN for Semi-Supervised Conditional Instance Synthesis and Classification (CVPR 2019 · similarity 0.91)
- MGAN: Training Generative Adversarial Nets with Multiple Generators (ICLR 2018 · similarity 0.90)
- Triangle Generative Adversarial Networks (NeurIPS 2017 · similarity 0.90)
- Improved Techniques for Training GANs (NeurIPS 2016 · similarity 0.90)
- Regularizing Discriminative Capability of CGANs for Semi-Supervised Generative Learning (CVPR 2020 · similarity 0.90)
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.