Research map / Image generation / GANs
Reciprocal GAN Through Characteristic Functions (RCF-GAN)
Shengxi Li, Zeyang Yu, Min Xiang, Danilo P. Mandic
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 reciprocal loss (lambda>0) in embedded space (ablation) | RCF-GAN · CelebA generation | FID: 59.39 -> 15.86 | Improved |
| add reciprocal loss (lambda>0) in embedded space (ablation) | RCF-GAN · CelebA reconstruction | FID: >100 -> 14.82 | Improved |
| add anchor design (ablation) | RCF-GAN · CelebA generation | FID: 17.80 -> 15.86 | Improved |
| add anchor design (ablation) | RCF-GAN · CelebA reconstruction | FID: >100 -> 14.82 | Improved |
| use learned t-net instead of fixed Gaussian t samples | RCF-GAN · CelebA generation | FID: 19.34 -> 15.86 | Improved |
| use learned t-net instead of fixed Gaussian t samples | RCF-GAN · LSUN Bedroom generation | FID: 38.16 -> 40.15 | Hurt |
| use learned t-net instead of fixed Gaussian t samples | RCF-GAN · CIFAR-10 generation | FID: 31.55 -> 31.21 | Improved |
| train generator solely via CF loss without critic latent comparison (ablation) | RCF-GAN · CelebA generation | FID: 15.86 -> 165 | Hurt |
| raise CF amplitude weight alpha from 0.5 to 0.75 (ablation) | RCF-GAN · CelebA generation | FID: 15.86 -> 13.84 | Improved |
| replace OCF-GAN-GP with RCF-GAN at 128x128 ResNet | OCF-GAN-GP · CelebA 128x128 | FID: 20.78 -> 10.71 | Improved |
| replace AGE with RCF-GAN for reconstruction | AGE · CIFAR-10 reconstruction | FID: 47.37 -> 28.70 | 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.
- MMD-GAN — MMD GAN: Towards Deeper Understanding of Moment Matching Network (NeurIPS 2017) · better on 3 of 3 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
- Reciprocal Adversarial Learning via Characteristic Functions (NeurIPS 2020 · similarity 0.99)
- Neural Characteristic Function Learning for Conditional Image Generation (ICCV 2023 · similarity 0.89)
- A Characteristic Function Approach to Deep Implicit Generative Modeling (CVPR 2020 · similarity 0.89)
- On the Statistical Properties of Generative Adversarial Models for Low Intrinsic Data Dimension (JMLR 2025 · similarity 0.88)
- PCF-GAN: generating sequential data via the characteristic function of measures on the path space (NeurIPS 2023 · similarity 0.87)
- Statistically Optimal Generative Modeling with Maximum Deviation from the Empirical Distribution (ICML 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.
