Research map / Image generation / GANs
Calibrating Energy-based 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 entropy regularization using the nearest-neighbor approximation instead of no regularization (ablation) | EGAN · Gaussian Mixture | KL(p_disc || p_data): 6.2159 -> 0.0901 | Helped |
| add entropy regularization using the nearest-neighbor approximation instead of no regularization (ablation) | EGAN · Biased Gaussian Mixture | KL(p_disc || p_data): 6.4969 -> 0.5199 | Helped |
| add entropy regularization using the nearest-neighbor approximation instead of no regularization (ablation) | EGAN · Two-spiral Gaussian Mixture | KL(p_disc || p_data): 3.9749 -> 0.4086 | Helped |
| use nearest-neighbor entropy approximation instead of variational inference (ablation) | EGAN-Ent · Gaussian Mixture | KL(p_disc || p_data): 0.7215 -> 0.0901 | Helped |
| use nearest-neighbor entropy approximation instead of variational inference (ablation) | EGAN-Ent · Biased Gaussian Mixture | KL(p_disc || p_data): 2.9894 -> 0.5199 | Helped |
| use nearest-neighbor entropy approximation instead of variational inference (ablation) | EGAN-Ent · Two-spiral Gaussian Mixture | KL(p_disc || p_data): 0.3061 -> 0.4086 | Hurt |
| use nearest-neighbor entropy approximation instead of variational inference (ablation) | EGAN-Ent · Biased Gaussian Mixture | KL(p_gen || p_data): 0.0303 -> 0.0334 | Hurt |
| use nearest-neighbor entropy approximation instead of variational inference (ablation) | EGAN-Ent · Gaussian Mixture | KL(p_gen || p_data): 0.0896 -> 0.0621 | Helped |
| use nearest-neighbor entropy approximation instead of variational inference (ablation) | EGAN-Ent · Two-spiral Gaussian Mixture | KL(p_gen || p_data): 0.4283 -> 0.4475 | Hurt |
Closest papers by meaning
- Energy-based Generative Adversarial Networks (ICLR 2017 · similarity 0.92)
- Improving Adversarial Energy-Based Model via Diffusion Process (ICML 2024 · similarity 0.89)
- A Unified Contrastive Energy-based Model for Understanding the Generative Ability of Adversarial Training (ICLR 2022 · similarity 0.89)
- Scalable Energy-Based Models via Adversarial Training: Unifying Discrimination and Generation (ICLR 2026 · similarity 0.89)
- Bounds all around: training energy-based models with bidirectional bounds (NeurIPS 2021 · similarity 0.89)
- On Stabilizing Generative Adversarial Training With Noise (CVPR 2019 · similarity 0.88)
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.