Research map / Image generation / GANs
Coupled Generative Adversarial Networks
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 |
|---|---|---|---|
| share more generative (decoder) layers across the two GANs (ablation) | CoGAN generative models · Task A digit-edge pairs | avg pixel agreement ratio: 0.894 (1 layer) to 0.951 (4 layers) | Helped |
| share more generative (decoder) layers across the two GANs (ablation) | CoGAN generative models · Task B digit-negative pairs | avg pixel agreement ratio: 0.932 (1 layer) to 0.979 (4 layers) | Helped |
| share discriminator last layers across the two GANs (ablation) | CoGAN discriminative models · Task A and Task B digit pairs | avg pixel agreement ratio | No clear effect |
| adding weight-sharing constraint (CoGAN vs unshared two GANs) (ablation) | CoGAN · digit pair generation | qualitative correspondence | Helped |
| CoGAN vs conditional GAN baseline (no pair supervision) | joint distribution learning on digits · Task B digit-negative pairs | avg pixel agreement ratio: 0.967 vs 0.778 | Helped |
| CoGAN vs conditional GAN baseline (no pair supervision) | joint distribution learning on digits · Task A digit-edge pairs | avg pixel agreement ratio: 0.952 vs 0.909 | Helped |
| CoGAN adaptation with shared discriminator layers for unsupervised domain adaptation | MNIST to USPS digit classification · UDA average accuracy | accuracy: 0.640 to 0.902 average | Helped |
| Cross-domain transformation via optimizing z through g1 then applying g2 | CoGAN image transformation · MNIST-like digit transformation | qualitative | Mixed |
Closest papers by meaning
- Constrained Generative Adversarial Networks for Interactive Image Generation (CVPR 2019 · similarity 0.89)
- AdaGAN: Boosting Generative Models (NeurIPS 2017 · similarity 0.88)
- JointGAN: Multi-Domain Joint Distribution Learning with Generative Adversarial Nets (ICML 2018 · similarity 0.88)
- Geometry-Consistent Generative Adversarial Networks for One-Sided Unsupervised Domain Mapping (CVPR 2019 · similarity 0.88)
- CcGAN: Continuous Conditional Generative Adversarial Networks for Image Generation (ICLR 2021 · similarity 0.88)
- CoopInit: Initializing Generative Adversarial Networks via Cooperative Learning (AAAI 2023 · 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.