Research map / Probabilistic and scientific modeling / Variational and generative sampling
Perturbative Black Box Variational Inference
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 |
|---|---|---|---|
| Use PBBVI (K = 3) instead of KLVI | Gaussian Process regression · GP regression synthetic data | average variance: 0.0176 -> 0.0355; analytic 0.0415 | Helped |
| Use PBBVI (K = 3) instead of KLVI | Gaussian Process classification · Crab test set | error rate: 0.22 -> 0.11 | Helped |
| Use PBBVI (K = 3) instead of KLVI | Gaussian Process classification · Pima test set | error rate: 0.245 -> 0.240 | Helped |
| Use PBBVI (K = 3) instead of KLVI | Gaussian Process classification · Heart test set | error rate: 0.148 -> 0.1333 | Helped |
| Use PBBVI (K = 3) instead of KLVI | Gaussian Process classification · Sonar test set | error rate: 0.212 -> 0.1731 | Helped |
| Use PBBVI (K = 3) instead of alpha-VI with α = 0.5 | Gaussian Process classification · Sonar | convergence speed: an order of magnitude faster | Helped |
| Use PBBVI (K = 3) instead of alpha-VI | Gaussian Process regression · synthetic data across latent dimensions | gradient sampling variance | Helped |
| Use PBBVI (K = 3) instead of KLVI | Variational Autoencoder · MNIST test set, smaller training subsets | predictive log-likelihood | Helped |
| Increase perturbation order from K = 1 to K = 3 (ablation) | PBBVI variational inference · general | bound tightness | Helped |
| Increase perturbation order (ablation) | PBBVI variational inference · general | gradient variance | Hurt |
| Use KLVI instead of PBBVI (ablation) | Variational Autoencoder · MNIST with larger training sets | predictive log-likelihood | No clear effect |
Closest papers by meaning
- Batch and match: black-box variational inference with a score-based divergence (ICML 2024 · similarity 0.91)
- Nearly Dimension-Independent Convergence of Mean-Field Black-Box Variational Inference (NeurIPS 2025 · similarity 0.90)
- Generalized Variational Inference via Optimal Transport (AAAI 2024 · similarity 0.89)
- EigenVI: score-based variational inference with orthogonal function expansions (NeurIPS 2024 · similarity 0.89)
- Efficient Mixture Learning in Black-Box Variational Inference (ICML 2024 · similarity 0.89)
- Forward $\chi^2$ Divergence Based Variational Importance Sampling (ICLR 2024 · similarity 0.89)
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.