Research map / Label-efficient and robust learning / Self-supervised and contrastive
Self-Supervised Enhancement of Latent Discovery in 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 |
|---|---|---|---|
| Apply SRE to SeFa directions | SeFa · CelebA-HQ | Rescoring: 0.64 -> 7.77 | Helped |
| Apply SRE to SeFa directions | SeFa · CelebA-HQ | Identity Preservation Accuracy: 0.61 -> 0.97 | Helped |
| Apply SRE to LD directions | LD · CelebA-HQ | Rescoring: 1.43 -> 3.73 | Helped |
| Apply SRE to LD directions | LD · CelebA-HQ | Identity Preservation Accuracy: 0.73 -> 0.94 | Helped |
| Apply SRE to SeFa directions | SeFa · 3DShapes | MIG: 0.22±0.01 -> 0.45±0.06 | Helped |
| Apply SRE to SeFa directions | SeFa · 3DShapes | Factor-VAE Score: 0.86 ± 0.01 -> 0.94±0.02 | Helped |
| Apply SRE to LD directions | LD · 3DShapes | MIG: 0.14±0.05 -> 0.21±0.05 | Helped |
| Apply SRE to LD directions | LD · 3DShapes | Factor-VAE Score: 0.78±0.06 -> 0.90±0.05 | Helped |
| Use random initialization instead of SeFa initialization (ablation) | SRE · 3DShapes | MIG: 0.45±0.06 -> 0.34 ± 0.04 | Hurt |
| Set ε range to (-3, 3) instead of (-1, 1) (ablation) | SRE · 3DShapes | MIG: 0.31 -> 0.26 | Hurt |
| Set ε range to (-10, 10) instead of (-1, 1) (ablation) | SRE · 3DShapes | MIG: 0.31 -> 0.18 | Hurt |
Closest papers by meaning
- Do Not Escape From the Manifold: Discovering the Local Coordinates on the Latent Space of GANs (ICLR 2022 · similarity 0.89)
- Unsupervised Discovery of Interpretable Directions in the GAN Latent Space (ICML 2020 · similarity 0.89)
- LARGE: Latent-Based Regression Through GAN Semantics (CVPR 2022 · similarity 0.89)
- InfoGAN-CR and ModelCentrality: Self-supervised Model Training and Selection for Disentangling GANs (ICML 2020 · similarity 0.89)
- Closed-Form Factorization of Latent Semantics in GANs (CVPR 2021 · similarity 0.89)
- Style Quantization for Data-Efficient GAN Training (CVPR 2025 · similarity 0.85)
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.