Research map / Probabilistic and scientific modeling / Variational and generative sampling
Quasi-Monte Carlo 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 |
|---|---|---|---|
| Replace MC sampling with RQMC sampling in the score-function estimator | MCVI · Hierarchical linear regression | convergence speed: up to three orders of magnitude | Helped |
| Replace MC sampling with RQMC sampling in the score-function estimator | MCVI · Hierarchical linear regression | gradient variance: up to three orders of magnitude | Helped |
| Replace 100 MC samples with 10 RQMC samples | MCVI with the reparameterization estimator · Hierarchical linear regression | gradient variance | No clear effect |
| Use RQMC sampling with 50 samples instead of MC sampling | MCVI with the reparameterization estimator · frisk | ELBO | Helped |
| Use RQMC sampling with 50 samples instead of MC or control variates | MCVI with the reparameterization estimator · frisk | gradient variance | Helped |
| Use RQMC sampling with 10 samples instead of MC sampling | MCVI with the reparameterization estimator · Bayesian neural network, wine dataset | ELBO | Helped |
| Use RQMC sampling with 50 samples instead of MC or control variates | MCVI with the reparameterization estimator · Bayesian neural network, wine dataset | convergence speed: one to three orders of magnitude | Helped |
| Use RQMC sampling with 50 samples instead of MC or control variates | MCVI with the reparameterization estimator · Bayesian neural network, wine dataset | gradient variance: one to three orders of magnitude | Helped |
| Increase the RQMC sample size over iterations while keeping a constant learning rate | SGD optimizing a strongly convex Monte Carlo objective · Two-dimensional normal target | asymptotic convergence rate | Helped |
| Replace ordinary MC sampling with RQMC sampling | MCVI with the reparameterization estimator · frisk, 10 samples | gradient variance | Hurt |
Closest papers by meaning
- A Divergence Bound for Hybrids of MCMC and Variational Inference and an Application to Langevin Dynamics and SGVI (ICML 2017 · similarity 0.91)
- Surrogate Likelihoods for Variational Annealed Importance Sampling (ICML 2022 · similarity 0.90)
- Structured Stochastic Gradient MCMC (ICML 2022 · similarity 0.90)
- A Contrastive Divergence for Combining Variational Inference and MCMC (ICML 2019 · similarity 0.89)
- Auxiliary Variational MCMC (ICLR 2019 · similarity 0.89)
- Online Variational Sequential Monte Carlo (ICML 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.