Research map / Optimization / Optimizers
Multilevel Monte Carlo Variational Inference
Masahiro Fujisawa, Issei Sato
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 MLMCVI instead of MCVI and QMCVI for gradient estimation | MLMCVI · HLR, BLR, BNN benchmarks | training ELBO convergence speed | Improved |
| use MLMCVI instead of MCVI and QMCVI | MLMCVI · HLR, BLR, BNN benchmarks | empirical gradient variance | Improved |
| use MLMCVI instead of MCVI and QMCVI | MLMCVI · HLR, BLR, BNN benchmarks | empirical SNR | Improved |
| use MLMCVI instead of baselines with a shared SGD and scheduler | MLMCVI · Bayesian logistic regression, Fashion-MNIST | test log-likelihood | Improved |
| use MLMCVI with reduced training data | MLMCVI · Bayesian logistic regression, Fashion-MNIST | test log-likelihood: 20% | Improved |
| vary scheduler hyperparameters (drop-rate r) and initial learning rate (ablation) | MLMCVI · HLR, BLR, BNN benchmarks | optimization performance | Mixed |
Closest papers by meaning
- Quasi-Monte Carlo Variational Inference (ICML 2018 · similarity 0.92)
- Sticking the Landing: Simple, Lower-Variance Gradient Estimators for Variational Inference (NeurIPS 2017 · similarity 0.91)
- VarGrad: A Low-Variance Gradient Estimator for Variational Inference (NeurIPS 2020 · similarity 0.91)
- Marginalized Stochastic Natural Gradients for Black-Box Variational Inference (ICML 2021 · similarity 0.90)
- Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm (NeurIPS 2016 · similarity 0.90)
- Stochastic variance-reduced Gaussian variational inference on the Bures-Wasserstein manifold (ICLR 2025 · 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.
