Research map / Image generation / GANs
Mode Regularized Generative Adversarial Networks
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 the proposed regularization | DCGAN · Compositional MNIST 1000 modes, Sets 1–4 | #Miss: 204.7 -> 32.1; 204.3 -> 71.5; 103.4 -> 42.7; 89.3 -> 31.6 | Helped |
| Add the proposed regularization | DCGAN · Compositional MNIST 1000 modes, Sets 1–4 | KL divergence: 77.9 -> 62.3; 60.2 -> 58.9; 75.9 -> 68.4; 77.8 -> 67.8 | Helped |
| Add the proposed regularizers | DCGAN · CelebA, σ=3.5 and σ=4.0, prior z=100 | missing-mode images: 5463 -> 754; 590 -> 42 | Helped |
| Add the proposed regularizers | DCGAN · CelebA, σ=3.5 and σ=4.0, prior z=200 | missing-mode images: 17089 -> 3644; 15832 -> 391 | Helped |
| Use manifold-diffusion training | DCGAN · CelebA, σ=3.5 and σ=4.0, prior z=200 | missing-mode images: 17089 -> 74; 15832 -> 13 | Helped |
| Replace VAEGAN with Reg-GAN | CelebA · CelebA, σ=3.5 and σ=4.0, prior z=100 | missing-mode images: 9720 -> 754; 5862 -> 42 | Helped |
| Replace VAEGAN with MDGAN | CelebA · CelebA, σ=3.5 and σ=4.0, prior z=200 | missing-mode images: 9720 -> 74; 5862 -> 13 | Helped |
| Increase DCGAN prior dimension from 100 to 200 | DCGAN · CelebA, σ=3.5 and σ=4.0 | missing-mode images: 5463 -> 17089; 590 -> 15832 | Hurt |
| Use Reg-GAN instead of DCGAN | CelebA · CelebA, σ=3.5 and σ=4.0 | missing-mode images: At prior z=100: 5463 -> 754 and 590 -> 42; at prior z=200: 17089 -> 3644 and 15832 -> 391 | Helped |
| Use MDGAN rather than the compared models | CelebA generators · CelebA generated samples | mode diversity | Helped |
Closest papers by meaning
- Stabilizing Training of Generative Adversarial Networks through Regularization (NeurIPS 2017 · similarity 0.93)
- A Large-Scale Study on Regularization and Normalization in GANs (ICML 2019 · similarity 0.92)
- Consistency Regularization for Generative Adversarial Networks (ICLR 2020 · similarity 0.91)
- No Modes Left Behind: Capturing the Data Distribution Effectively Using GANs (AAAI 2018 · similarity 0.91)
- Improving Generalization and Stability of Generative Adversarial Networks (ICLR 2019 · similarity 0.90)
- Regularizing Generative Adversarial Networks Under Limited Data (CVPR 2021 · similarity 0.90)
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.