Research map / Optimization / Optimizers
Quasi-Monte Carlo Quasi-Newton in Variational Bayes
Sifan Liu, Art B. Owen
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 |
|---|---|---|---|
| replace MC with RQMC (scrambled Sobol) sampling in AdaGrad | AdaGrad on Bayesian linear regression VB · Bayesian linear regression (N=300, d=100) | log2 error vs n | Improved |
| replace MC with RQMC in AdaGrad | AdaGrad · Bayesian logistic regression VB (N=30, d=100) | ELBO vs wall clock | Improved |
| use L-BFGS instead of AdaGrad | MC and RQMC sampling · Bayesian logistic regression VB (N=30, d=100) | ELBO convergence speed | Improved |
| replace MC with RQMC sampling in L-BFGS (8 gradient samples) | L-BFGS · Crossed random effects VB (ng=8) | ELBO fluctuation | Improved |
| replace MC with RQMC sampling in BFGS (128 gradient samples) | BFGS · Crossed random effects VB (ng=128) | ELBO | Improved |
| use BFGS instead of AdaGrad | MC and RQMC sampling · Crossed random effects VB | time to optimum / ELBO | Improved |
| replace MC with RQMC in BFGS | BFGS · VAE on MNIST | ELBO | Improved |
| use BFGS instead of Adam | VAE with MC and RQMC · VAE on MNIST | convergence speed | Improved |
| go deeper into one RQMC sequence rather than fresh randomization each step | RQMC-SQN · VB examples | optimization performance | No clear effect |
Closest papers by meaning
- Quasi-Monte Carlo Variational Inference (ICML 2018 · similarity 0.93)
- Stochastic Quasi-Newton Langevin Monte Carlo (ICML 2016 · similarity 0.90)
- Langevin Quasi-Monte Carlo (NeurIPS 2023 · similarity 0.89)
- Quasi-Monte Carlo Features for Kernel Approximation (ICML 2024 · similarity 0.88)
- Zeroth-Order Sampling Methods for Non-Log-Concave Distributions: Alleviating Metastability by Denoising Diffusion (NeurIPS 2024 · similarity 0.86)
- Accelerating Look-ahead in Bayesian Optimization: Multilevel Monte Carlo is All you Need (ICML 2024 · 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.
