Research map / Image generation / GANs
Interpretable 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 |
|---|---|---|---|
| Add the interpretability loss to StyleGAN | StyleGAN · CelebA-HQ | FID: 12.86 -> 18.81 | Hurt |
| Add the interpretability loss to BigGAN | BigGAN · FFHQ | FID: 41.81 -> 56.74 | Hurt |
| Use interpretable intermediate features for localized concept editing | Interpretable StyleGAN · Mouth editing | human evaluation score (%): 56.00 -> 83.60 | Helped |
| Use interpretable intermediate features for localized concept editing | Interpretable BigGAN · Eyes editing | human evaluation score (%): 45.40 -> 82.10 | Helped |
| Use interpretable intermediate features for localized concept editing | Interpretable StyleGAN · Chin editing | locality metric: 0.0576 -> 0.0163 | Helped |
| Use interpretable intermediate features for localized concept editing | Interpretable StyleGAN · Face swapping | face verification accuracy (%): 89.20 -> 90.25 | Helped |
| Perform Langevin dynamics on generated images (ablation) | Interpretable StyleGAN · CelebA-HQ | FID: 18.81 -> 19.42 | Hurt |
| Perform Langevin dynamics on generated images (ablation) | Interpretable BigGAN · FFHQ | FID: 56.74 -> 57.72 | Hurt |
| Increase the number of visual-concept groups C (ablation) | Interpretable GAN · Generated images | visual concept detail | Helped |
Closest papers by meaning
- Restyling Unsupervised Concept Based Interpretable Networks with Generative Models (ICLR 2025 · similarity 0.88)
- Navigating the GAN Parameter Space for Semantic Image Editing (CVPR 2021 · similarity 0.88)
- Linear Semantics in Generative Adversarial Networks (CVPR 2021 · similarity 0.88)
- GANSpace: Discovering Interpretable GAN Controls (NeurIPS 2020 · similarity 0.88)
- Network Dissection: Quantifying Interpretability of Deep Visual Representations (CVPR 2017 · similarity 0.88)
- Interpretable Generative Models through Post-hoc Concept Bottlenecks (CVPR 2025 · similarity 0.87)
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.