Research map / Image generation / GANs
Learning Fixed Points in Generative Adversarial Networks: From Image-to-Image Translation to Disease Detection and Localization
Md Mahfuzur Rahman Siddiquee, Zongwei Zhou, Nima Tajbakhsh, Ruibin Feng, Michael B. Gotway, Yoshua Bengio, Jianming Liang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 fixed-point (same-domain) translation learning to StarGAN (ablation) | StarGAN · BRATS 2013 image-level detection | AUC: 0.4611 -> 0.9980 | Improved |
| add delta map only to StarGAN (ablation) | StarGAN · BRATS 2013 image-level detection | AUC: 0.4611 -> 0.5246 | No clear effect |
| add delta map to fixed-point translation (ablation) | Fixed-Point GAN · BRATS 2013 image-level detection | AUC: 0.9980 -> 0.9831 | Hurt |
| add delta map only to StarGAN (ablation) | StarGAN · PE image-level detection | AUC: 0.8832 -> 0.8603 | Hurt |
| add fixed-point translation learning to StarGAN (ablation) | StarGAN · PE image-level detection | AUC: 0.8832 -> 0.9216 | Improved |
| add delta map to fixed-point translation (ablation) | Fixed-Point GAN · PE image-level detection | AUC: 0.9216 -> 0.9668 | Improved |
| add fixed-point translation learning to StarGAN (ablation) | StarGAN · BRATS 2013 lesion localization | sensitivity at 1 FP: 13.6% -> 81.2% | Improved |
| add delta map to fixed-point translation (ablation) | Fixed-Point GAN · BRATS 2013 lesion localization | sensitivity at 1 FP: 81.2% -> 84.5% | Improved |
| add fixed-point translation learning to StarGAN (ablation) | StarGAN · PE localization | sensitivity at 1 FP: 88.9% -> 94.4% | Improved |
| add delta map to fixed-point translation (ablation) | Fixed-Point GAN · PE localization | sensitivity at 1 FP: 94.4% -> 97.2% | Improved |
| replace StarGAN with Fixed-Point GAN | CelebA same-domain translation · CelebA same-domain translation | image-level L1 distance: 2.40 -> 0.36 | Improved |
| replace StarGAN with Fixed-Point GAN | CelebA cross-domain translation · CelebA cross-domain translation | classifier accuracy: 90.82% -> 92.31% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- StarGAN — StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation (CVPR 2018) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TraVeLGAN: Image-To-Image Translation by Transformation Vector Learning (CVPR 2019 · similarity 0.88)
- T-GD: Transferable GAN-generated Images Detection Framework (ICML 2020 · similarity 0.87)
- Towards Discovery and Attribution of Open-World GAN Generated Images (ICCV 2021 · similarity 0.87)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 2024 · similarity 0.86)
- MSG-GAN: Multi-Scale Gradients for Generative Adversarial Networks (CVPR 2020 · similarity 0.86)
- RG-GAN: Dynamic Regenerative Pruning for Data-Efficient Generative Adversarial Networks (AAAI 2024 · similarity 0.86)
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.
