Research map / Label-efficient and robust learning / Semi-supervised learning
Improved Techniques for Training GANs
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 |
|---|---|---|---|
| Add minibatch features (minibatch discrimination) to the discriminator (ablation) | GAN generating CIFAR-10 samples · CIFAR-10 Inception score | Inception score: 3.87 -> 8.09 (full methods vs -MBF) | Helped |
| Add labels (semi-supervised class information) to GAN training (ablation) | GAN generating CIFAR-10 samples · CIFAR-10 Inception score | Inception score: 4.36 -> 8.09 (vs -L) | Helped |
| Add one-sided label smoothing to the discriminator (ablation) | GAN generating CIFAR-10 samples · CIFAR-10 Inception score | Inception score: 6.83 -> 8.09 | Helped |
| Use virtual batch normalization in generator instead of batch normalization (ablation) | GAN generating CIFAR-10 samples · CIFAR-10 Inception score | Inception score: 7.54 -> 8.09 | Helped |
| Add historical averaging (with labels removed) (ablation) | GAN generating CIFAR-10 samples without labels · CIFAR-10 Inception score | Inception score: 4.36 -> 6.86 | Helped |
| Add feature matching objective for semi-supervised learning on MNIST/CIFAR-10/SVHN | GAN-based semi-supervised classifier · MNIST, CIFAR-10 semi-supervised classification | test error | Helped |
| Use minibatch discrimination instead of feature matching for generator training | GAN-based semi-supervised classifier · MNIST semi-supervised | classification quality | Hurt |
| Add minibatch discrimination to generator training | GAN generating MNIST samples · MNIST sample quality (MTurk) | human discrimination rate: annotators correct 52.4% (near chance) | Helped |
Closest papers by meaning
- Sequential Training of GANs Against GAN-Classifiers Reveals Correlated "Knowledge Gaps" Present Among Independently Trained GAN Instances (CVPR 2023 · similarity 0.91)
- SGAN: An Alternative Training of Generative Adversarial Networks (CVPR 2018 · similarity 0.91)
- A Large-Scale Study on Regularization and Normalization in GANs (ICML 2019 · similarity 0.91)
- Training Generative Adversarial Networks with Limited Data (NeurIPS 2020 · similarity 0.91)
- Large Scale GAN Training for High Fidelity Natural Image Synthesis (ICLR 2019 · similarity 0.91)
- Progressive Growing of GANs for Improved Quality, Stability, and Variation (ICLR 2018 · similarity 0.91)
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.