Research map / Label-efficient and robust learning / Recognition and long-tail
Rebooting ACGAN: Auxiliary Classifier GANs with Stable Training
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 |
|---|---|---|---|
| replace ACGAN conditioning with ReACGAN | BigGAN · Tiny-ImageNet | FID: 28.55 -> 27.10 | Helped |
| replace ACGAN conditioning with ReACGAN | BigGAN · ImageNet (B.S.=256) | FID: 16.36 -> 13.98 | Helped |
| add feature and weight normalization (ablation) | ACGAN · Tiny-ImageNet | FID: 96.04 -> 30.33 | Helped |
| add data-to-data comparisons (ablation) | ACGAN + normalization · Tiny-ImageNet | FID: 30.33 -> 28.71 | Helped |
| add easy positive and negative sample suppression (ablation) | ACGAN with normalization and data-to-data comparisons · Tiny-ImageNet | FID: 28.71 -> 27.10 | Helped |
| remove data-to-data comparisons (ablation) | ReACGAN · Tiny-ImageNet | FID: 27.10 -> 30.79 | Hurt |
| add feature and weight normalization (ablation) | ACGAN · CUB200 | FID: 60.73 -> 25.54 | Helped |
| add easy positive and negative sample suppression (ablation) | ACGAN with normalization and data-to-data comparisons · CUB200 | FID: 25.12 -> 15.40 | Helped |
| add differentiable augmentation | ReACGAN · CIFAR10 | FID: 7.88 -> 6.79 | Helped |
| add D2D-CE conditioning | StyleGAN2 · CIFAR10 | FID: 3.88 -> 3.34 | Helped |
| retain all negative-sample similarities (masking p=1.0) (ablation) | ReACGAN D2D-CE · CIFAR10 | FID: 7.88 -> 11.15 | Hurt |
| add TAC (ablation) | ReACGAN · CIFAR10 | FID: 7.88 -> 7.94 | Hurt |
Closest papers by meaning
- A Unified View of cGANs with and without Classifiers (NeurIPS 2021 · similarity 0.93)
- Twin Auxilary Classifiers GAN (NeurIPS 2019 · similarity 0.91)
- Conditional GANs with Auxiliary Discriminative Classifier (ICML 2022 · similarity 0.90)
- Turning Waste into Wealth: Leveraging Low-Quality Samples for Enhancing Continuous Conditional Generative Adversarial Networks (AAAI 2024 · similarity 0.88)
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 2024 · similarity 0.88)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.88)
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.