Research map / Image generation / GANs
Structure-preserving GANs
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 CNN generator and discriminator with equivariant generator and invariant discriminator (C4) | RA-GAN · RotMNIST 1% | FID: 295 -> 98 | Helped |
| replace CNN generator and discriminator with equivariant generator and invariant discriminator (C4) | Dα=2Γ-GAN · RotMNIST 1% | FID: 280 -> 149 | Helped |
| add an equivariant generator while retaining a CNN discriminator (ablation) | RA-GAN · RotMNIST 1% | FID: 295 -> 389 | Hurt |
| add an invariant discriminator while retaining a CNN generator (ablation) | RA-GAN · RotMNIST 1% | FID: 295 -> 223 | Helped |
| replace full equivariant generator with the incomplete equivariant generator (ablation) | RA-GAN with invariant discriminator · RotMNIST 1% | FID: 98 -> 173 | Hurt |
| increase symmetry group from C4 to C8 (ablation) | RA-GAN with equivariant generator and invariant discriminator · RotMNIST 1% | FID: 98 -> 123 | Hurt |
| increase symmetry group from C4 to C8 (ablation) | RA-GAN with equivariant generator and invariant discriminator · RotMNIST 5% | FID: 78 -> 52 | Helped |
| add an equivariant generator while retaining a CNN discriminator (ablation) | Dα=2Γ-GAN · RotMNIST 1% | FID: 280 -> 253 | Helped |
| add an invariant discriminator while retaining a CNN generator (ablation) | Dα=2Γ-GAN · RotMNIST 1% | FID: 280 -> 330 | Hurt |
| replace full equivariant generator with the incomplete equivariant generator (ablation) | Dα=2Γ-GAN with invariant discriminator · RotMNIST 1% | FID: 149 -> 273 | Hurt |
| increase symmetry group from C4 to C8 (ablation) | Dα=2Γ-GAN with equivariant generator and invariant discriminator · RotMNIST 1% | FID: 149 -> 122 | Helped |
| use invariant discriminator without ensuring generator output is invariant (ablation) | GANs on the C4-symmetric toy mixture · Toy example | mode coverage | Hurt |
Closest papers by meaning
- Dual Discriminator Generative Adversarial Nets (NeurIPS 2017 · similarity 0.91)
- Mixed batches and symmetric discriminators for GAN training (ICML 2018 · similarity 0.90)
- Improving GAN with Neighbors Embedding and Gradient Matching (AAAI 2019 · similarity 0.90)
- Group Equivariant Generative Adversarial Networks (ICLR 2021 · similarity 0.90)
- A Convex Duality Framework for GANs (NeurIPS 2018 · similarity 0.90)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.89)
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.