Research map / Continual learning / Online and replay-based
CAM-GAN: Continual Adaptation Modules for Generative Adversarial Networks
Sakshi Varshney, Vinay Kumar Verma, Srijith P K, Lawrence Carin, Piyush Rai
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 |
|---|---|---|---|
| Use CAM-GAN instead of GAN-Memory | continual unconditional generation · Cathedral | FID: 15.6 -> 9.52 | Improved |
| Use CAM-GAN instead of GAN-Memory | continual unconditional generation · X-Ray | FID: 17.18 -> 14.24 | Improved |
| Use CAM-GAN instead of GAN-Memory | continual unconditional generation · Flower | FID: 14.8 -> 23.0 | Hurt |
| Use CAM-GAN instead of GAN-Memory | continual unconditional generation · Anime | FID: 13.29 -> 13.1 | Improved |
| Initialize adapters from the most similar task using FIM-based task similarity (ablation) | CAM-GAN · Cat | FID: 18.2 -> 16.4 | Improved |
| Initialize adapters from the most similar task using FIM-based task similarity (ablation) | CAM-GAN · Brain-MRI | FID: 15.8 -> 15.0 | Improved |
| Initialize adapters from the most similar task using FIM-based task similarity (ablation) | CAM-GAN · Anime | FID: 13.3 -> 10.9 | Improved |
| Reduce adapter parameter growth to 50% while using task-similarity initialization (ablation) | CAM-GAN · Cat | FID: 16.4 -> 18.8 | Hurt |
| Reduce adapter parameter growth to 50% while using task-similarity initialization (ablation) | CAM-GAN · Brain-MRI | FID: 15.0 -> 15.9 | Hurt |
| Reduce adapter parameter growth to 50% while using task-similarity initialization (ablation) | CAM-GAN · Anime | FID: 10.9 -> 13.8 | Hurt |
| Add the 1x1 groupwise convolution adapter alongside the 3x3 groupwise adapter (ablation) | CAM-GAN · Cat and Brain-MRI | FID | Improved |
| Add residual bias in transformed feature space (ablation) | CAM-GAN generator training · CelebA | training stability: 125K iterations | Improved |
Later papers that compared with this method
1 later accepted paper reports CAM-GAN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Continual Unsupervised Generative Modeling (TPAMI 2025 · better on 1 of 1 shared result)
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
- Lifelong GAN: Continual Learning for Conditional Image Generation (ICCV 2019 · similarity 0.89)
- Efficient Feature Transformations for Discriminative and Generative Continual Learning (CVPR 2021 · similarity 0.88)
- Recall-Oriented Continual Learning with Generative Adversarial Meta-Model (AAAI 2024 · similarity 0.87)
- Resurrecting Old Classes with New Data for Exemplar-Free Continual Learning (CVPR 2024 · similarity 0.86)
- Consistency-GAN: Training GANs with Consistency Model (AAAI 2024 · similarity 0.85)
- RG-GAN: Dynamic Regenerative Pruning for Data-Efficient Generative Adversarial Networks (AAAI 2024 · 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.
