Research map / Image generation / GANs
Sub-GAN: An Unsupervised Generative Model via Subspaces
Jie Liang, Jufeng Yang, Hsin-Ying Lee, Kai Wang, Ming-Hsuan Yang
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 refinement of cluster assignment in the discriminator (ablation) | Sub-GAN · MNIST clustering (K=10), last epoch | clustering accuracy (%): 83.87 -> 85.32 | Improved |
| add refinement of cluster assignment in the discriminator (ablation) | Sub-GAN · MNIST clustering (K=10), 1st epoch | clustering accuracy (%): 75.23 -> 77.12 | Improved |
| add refinement of cluster assignment in the discriminator (ablation) | Sub-GAN · MNIST clustering (K=10), 40th epoch | clustering accuracy (%): 83.11 -> 84.24 | Improved |
| joint training of generator, discriminator and clusterer after SMR initialization | Sub-GAN · MNIST clustering (K=10) | clustering accuracy (%): 73.39 -> 85.32 | Improved |
| joint training of generator, discriminator and clusterer after SMR initialization | Sub-GAN · CIFAR clustering (K=10) | clustering accuracy (%): 56.24 -> 78.95 | Improved |
| use subspace-conditioned generation (Sub-GAN) instead of InfoGAN | InfoGAN · MNIST (K=10) | diversity score: 2.11 -> 2.36 | Improved |
| use subspace-conditioned generation (Sub-GAN) instead of InfoGAN | InfoGAN · CIFAR (K=10) | diversity score: 2.48 -> 2.72 | Improved |
| use Sub-GAN instead of IGAN | IGAN · CIFAR (K=10) | diversity score: 2.20 -> 2.72 | Improved |
| increase number of clusters K from 10 to 20 (ablation) | Sub-GAN · MNIST clustering | adjusted clustering accuracy (%): 85.32 -> 90.81 | Improved |
| increase number of clusters K from 10 to 20 (ablation) | Sub-GAN · CIFAR clustering | adjusted clustering accuracy (%): 78.95 -> 82.44 | Improved |
| increase number of clusters K from 10 to 20 (ablation) | DEC · MNIST clustering | adjusted clustering accuracy (%): 84.30 -> 83.02 | Hurt |
Closest papers by meaning
- Low-Rank Subspaces in GANs (NeurIPS 2021 · similarity 0.90)
- DeLiGAN : Generative Adversarial Networks for Diverse and Limited Data (CVPR 2017 · similarity 0.90)
- Exploring the Precise Dynamics of Single-Layer GAN Models: Leveraging Multi-Feature Discriminators for High-Dimensional Subspace Learning (NeurIPS 2024 · similarity 0.89)
- A Geometric Analysis of Deep Generative Image Models and Its Applications (ICLR 2021 · similarity 0.89)
- Train Sparsely, Generate Densely: Memory-Efficient Unsupervised Training of High-Resolution Temporal GAN (IJCV 2020 · similarity 0.88)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.87)
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.
