Research map / Learning theory and algorithms / Kernels and regression theory
PAC-Bayesian Generalization Bounds for Adversarial Generative Models
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 |
|---|---|---|---|
| increase prior std sigma0 from 1e-3 to 1e-2 (too-large prior variance) (ablation) | probabilistic WGAN on MNIST · MNIST | FID (lower is better): 112.51 -> 189.30 | Hurt |
| sigma0 in range 1e-7 to 1e-3 on MNIST (ablation) | probabilistic WGAN · MNIST | FID (lower is better): 106.59 to 113.16 | No clear effect |
| sigma0=0.1 on synthetic datasets makes models fail to learn the data distribution (larger empirical and true risks) (ablation) | probabilistic WGAN on 8-Gaussian ring and 25-Gaussian grid · synthetic datasets | risk (negative critic loss) and generated sample quality | Hurt |
| learn prior mean on a portion of training data rather than fixing it to zero or random | PAC-Bayes bound optimization for neural networks · synthetic datasets | bound tightness / balance of objective | Helped |
Closest papers by meaning
- Generative Modeling Using the Sliced Wasserstein Distance (CVPR 2018 · similarity 0.89)
- Asymptotic Guarantees for Learning Generative Models with the Sliced-Wasserstein Distance (NeurIPS 2019 · similarity 0.89)
- PAC-Bayesian Spectrally-Normalized Bounds for Adversarially Robust Generalization (NeurIPS 2023 · similarity 0.88)
- Asymptotic Guarantees for Generative Modeling Based on the Smooth Wasserstein Distance (NeurIPS 2020 · similarity 0.88)
- Statistically Optimal Generative Modeling with Maximum Deviation from the Empirical Distribution (ICML 2024 · similarity 0.88)
- Universal generalization guarantees for Wasserstein distributionally robust models (ICLR 2025 · 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.