Research map / Image generation / GANs
An Efficient Explorative Sampling Considering the Generative Boundaries of Deep Generative Neural Networks
Giyoung Jeon, Haedong Jeong, Jaesik Choi
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 |
|---|---|---|---|
| Sample within the model-defined generative region using E-GBAS instead of ε-based sampling | DCGAN · MNIST, layers 1–4 | standard deviation: 0.0819 -> 0.0781; 0.0711 -> 0.0694; 0.0718 -> 0.0675; 0.0343 -> 0.0323 | Improved |
| Sample within the model-defined generative region using E-GBAS instead of ε-based sampling | PGGAN · LSUN, layers 2, 4, and 6 | standard deviation: 0.4641 -> 0.3116; 0.4322 -> 0.2558; 0.3365 -> 0.1748 | Improved |
| Sample within the model-defined generative region using E-GBAS instead of ε-based sampling | PGGAN · celebA, layers 2, 4, and 6 | standard deviation: 0.4859 -> 0.2980; 0.4799 -> 0.2789; 0.3384 -> 0.1446 | Improved |
| Use E-GBAS rather than ε-based sampling | DCGAN · MNIST, discriminator layers 1 and 4 | cosine similarity: 0.722 -> 0.747; 0.991 -> 0.992 | Improved |
| Use E-GBAS rather than ε-based sampling | PGGAN · LSUN, discriminator layers 1 and 4 | cosine similarity: 0.551 -> 0.578; 0.920 -> 1.000 | Improved |
| Use E-GBAS rather than ε-based sampling | PGGAN · celebA, discriminator layers 1 and 4 | cosine similarity: 0.684 -> 0.702; 0.965 -> 0.970 | Improved |
Closest papers by meaning
- Polarity Sampling: Quality and Diversity Control of Pre-Trained Generative Networks via Singular Values (CVPR 2022 · similarity 0.89)
- Plug & Play Generative Networks: Conditional Iterative Generation of Images in Latent Space (CVPR 2017 · similarity 0.88)
- Boundary Seeking GANs (ICLR 2018 · similarity 0.86)
- Large Scale GAN Training for High Fidelity Natural Image Synthesis (ICLR 2019 · similarity 0.86)
- DeLiGAN : Generative Adversarial Networks for Diverse and Limited Data (CVPR 2017 · similarity 0.86)
- Consistency-GAN: Training GANs with Consistency Model (AAAI 2024 · similarity 0.85)
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.
