Research map / Probabilistic and scientific modeling / Variational and generative sampling
Boosting 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 boosting BBVI with Laplace base distributions instead of baseline BBVI with Gaussian variational distributions | Bayesian logistic regression · ChemReact | AUROC | Helped |
| Use boosting BBVI with Laplace base distributions instead of baseline BBVI | Bayesian logistic regression · ChemReact | standard deviation across replications | Helped |
| Use boosting BBVI instead of baseline BBVI | Bayesian logistic regression · eICU mortality prediction | AUROC | Helped |
| Use boosting BBVI instead of baseline BBVI | Bayesian logistic regression · eICU mortality prediction | standard deviation across replications | Helped |
| Use boosting BBVI instead of BBVI | Bayesian matrix factorization · CBCL Face, D=3 | mean-squared error | Helped |
| Use boosting BBVI instead of BBVI | Bayesian matrix factorization · CBCL Face, D=5 | mean-squared error | Helped |
| Use boosting BBVI instead of BBVI | Bayesian matrix factorization · CBCL Face, D=10 | mean-squared error | Helped |
| Use boosting BBVI instead of BBVI | Bayesian matrix factorization · CBCL Face, D=3 | held-out test log-likelihood | Helped |
| Use boosting BBVI instead of BBVI | Bayesian matrix factorization · CBCL Face | variance across replications | Helped |
| Increase latent dimension from D=3 to D=5 or D=10 (ablation) | Bayesian matrix factorization · CBCL Face | mean-squared error | No clear effect |
| Increase latent dimension from D=3 to D=5 or D=10 (ablation) | Bayesian matrix factorization · CBCL Face | held-out test log-likelihood | No clear effect |
| Use fully corrective Frank-Wolfe instead of fixed-step Frank-Wolfe (ablation) | Boosting BBVI · Synthetic bimodal Gaussian posterior | fit to target distribution | Helped |
Closest papers by meaning
- Variational Boosting: Iteratively Refining Posterior Approximations (ICML 2017 · similarity 0.92)
- Theoretical Guarantees for Variational Inference with Fixed-Variance Mixture of Gaussians (ICML 2024 · similarity 0.89)
- Variational Inference with Mixtures of Isotropic Gaussians (NeurIPS 2025 · similarity 0.89)
- Efficient Mixture Learning in Black-Box Variational Inference (ICML 2024 · similarity 0.89)
- Batch and match: black-box variational inference with a score-based divergence (ICML 2024 · similarity 0.88)
- EigenVI: score-based variational inference with orthogonal function expansions (NeurIPS 2024 · 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.