Research map / Label-efficient and robust learning / Recognition and long-tail
Adversarial Fisher Vectors for Unsupervised Representation Learning
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 |
|---|---|---|---|
| Use AFVs instead of pooled discriminator features (ablation) | GAN discriminator features · CIFAR10 | classification accuracy: 65.3 -> 86.2 | Helped |
| Add data augmentation during GAN training (ablation) | AFV-128-50000 · CIFAR10 | classification accuracy: 86.2 -> 87.1 | Helped |
| Increase model channel size to 256 (ablation) | AFV-128-50000 + augment · CIFAR10 | classification accuracy: 87.1 -> 88.5 | Helped |
| Add CIFAR100 training data (ablation) | AFV-256-50000 + augment · CIFAR10 | classification accuracy: 88.5 -> 89.1 | Helped |
| Add CIFAR100 training data (ablation) | AFV-256-50000 + augment · CIFAR100 | classification accuracy | Helped |
| Use AFV as the perceptual similarity metric | ImageNet-trained GAN · BAPPS traditional distortions | 2AFC: 70.6 -> 74.9 | Helped |
| Use AFV as the perceptual similarity metric | ImageNet-trained GAN · BAPPS CNN distortions | 2AFC: 83.1 -> 80.2 | Hurt |
| Use AFV as the perceptual similarity metric | ImageNet-trained GAN · BAPPS average | Avg 2AFC: 78.0 -> 77.6 | Hurt |
| Use AFV as the perceptual similarity metric | ImageNet-trained GAN · BAPPS traditional distortions | JND: 44.0 -> 55.7 | Helped |
| Use AFV as the perceptual similarity metric | ImageNet-trained GAN · BAPPS CNN distortions | JND: 67.6 -> 62.0 | Hurt |
| Use AFV as the perceptual similarity metric | ImageNet-trained GAN · BAPPS average | Avg JND: 55.6 -> 58.9 | Helped |
| Constrain generator updates to be local with the MCMC-inspired objective (ablation) | GAN generator · ImageNet 64x64 | Inception Score | Helped |
Closest papers by meaning
- Fisher GAN (NeurIPS 2017 · similarity 0.90)
- Scalable Energy-Based Models via Adversarial Training: Unifying Discrimination and Generation (ICLR 2026 · similarity 0.88)
- Calibrating Energy-based Generative Adversarial Networks (ICLR 2017 · similarity 0.87)
- Energy-based Generative Adversarial Networks (ICLR 2017 · similarity 0.87)
- Structure-preserving GANs (ICML 2022 · similarity 0.87)
- Improving Adversarial Energy-Based Model via Diffusion Process (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.