Research map / Probabilistic and scientific modeling / Variational and generative sampling
Hamiltonian Monte Carlo with Energy Conserving Subsampling
Khue-Dung Dang, Matias Quiroz, Robert Kohn, Minh-Ngoc Tran, Mattias Villani
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 perturbed or signed HMC-ECS instead of SG-HMC with control variates | logistic regression · HIGGS (logistic regression) | relative computational time: median RCT 1.15 (signed HMC-ECS) vs 2.97 (SG-HMC) | Improved |
| use HMC-ECS instead of SGLD | logistic regression · HIGGS (logistic regression) | posterior accuracy: median RCT 1.15 vs 12.46; SGLD biased | Improved |
| add control variates to SG-HMC (ablation) | SG-HMC · HIGGS (logistic regression) | posterior accuracy | Improved |
| add control variates to SGLD (ablation) | SGLD · HIGGS (logistic regression) | posterior accuracy | No clear effect |
| use HMC-ECS instead of SG-HMC with L=60 | spline logistic model · Bankruptcy (spline logistic model) | relative computational time: median RCT 1.6 vs 9.5 | Improved |
| reduce SG-HMC leapfrog steps from L=60 to L=7 (ablation) | SG-HMC · Bankruptcy (spline logistic model) | relative computational time and accuracy: median RCT 9.5 -> 100.2 | Hurt |
| use signed instead of perturbed HMC-ECS | HMC-ECS · Bankruptcy (spline logistic model) | inefficiency factor: IF 2.202 -> 2.31 | Hurt |
| use first-order instead of second-order Taylor control variates (ablation) | HMC-ECS · Limitations experiment (control variate order) | sampling performance | No clear effect |
| use zeroth-order instead of higher-order control variates (ablation) | HMC-ECS · Limitations experiment (control variate order) | variance of log-likelihood estimator | Hurt |
Closest papers by meaning
- Hamiltonian Monte Carlo for efficient Gaussian sampling: long and random steps (JMLR 2024 · similarity 0.93)
- A Gradient Based Strategy for Hamiltonian Monte Carlo Hyperparameter Optimization (ICML 2021 · similarity 0.92)
- On the Convergence of Hamiltonian Monte Carlo with Stochastic Gradients (ICML 2021 · similarity 0.92)
- Entropy-based adaptive Hamiltonian Monte Carlo (NeurIPS 2021 · similarity 0.91)
- Practical and Scalable Hamiltonian Monte Carlo Without the Metropolis Test (ICML 2026 · similarity 0.91)
- Metropolis Adjusted Microcanonical Hamiltonian Monte Carlo (NeurIPS 2025 · similarity 0.90)
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.
