Research map / Optimization / Optimizers
A Hybrid Stochastic Gradient Hamiltonian Monte Carlo Method
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 the hybrid stochastic-gradient estimator with a single sample per iteration | HSG-HMC · Nonconvex potentials, 2-Wasserstein ε-accuracy | overall sample complexity: ˜O(ϵ−4µ−2∗) | Helped |
| Use HSG-HMC instead of SRVR-HMC | SG-MCMC methods · Nonconvex potentials, 2-Wasserstein ε-accuracy | hyperparameters: 4 -> 1 | Helped |
| Use one sample per iteration rather than SRVR-HMC's periodically large mini-batches | HSG-HMC · Nonconvex potentials, 2-Wasserstein ε-accuracy | mini-batch size: ˜O(ϵ−4µ−1∗) -> 1 | Helped |
| Use HSG-HMC instead of SRVR-HMC | Gaussian Mixture Density · Gaussian Mixture Density | W2: 5.33 ∗ 10−3 -> 5.13 ∗ 10−3 | Helped |
| Use HSG-HMC instead of SVRG-LD | Gaussian Mixture Density · Gaussian Mixture Density | W2: 7.82 ∗ 10−3 -> 5.13 ∗ 10−3 | Helped |
| Use variance-reduced methods rather than methods without variance reduction | SG-MCMC sampling · Bayesian Neural Network, MNIST | test negative log-likelihood and accuracy | Helped |
| Use underdamped Langevin dynamics rather than overdamped Langevin dynamics | SG-MCMC sampling · Bayesian Neural Network, MNIST | test negative log-likelihood and accuracy | Helped |
| Use HSG-HMC instead of other compared methods | Bayesian Logistic Regression · a9a, mushrooms, phishing, pima, a3a, ijcnn | test negative log-likelihood | Helped |
| Use HSG-HMC instead of other compared methods | Bayesian Neural Network · MNIST | test negative log-likelihood and accuracy | Helped |
Closest papers by meaning
- Stochastic Gradient Hamiltonian Monte Carlo Methods with Recursive Variance Reduction (NeurIPS 2019 · similarity 0.94)
- On the Convergence of Hamiltonian Monte Carlo with Stochastic Gradients (ICML 2021 · similarity 0.94)
- Stochastic Quasi-Newton Langevin Monte Carlo (ICML 2016 · similarity 0.92)
- Stochastic Variance-Reduced Hamilton Monte Carlo Methods (ICML 2018 · similarity 0.92)
- Stochastic Gradient Richardson-Romberg Markov Chain Monte Carlo (NeurIPS 2016 · similarity 0.91)
- Parameter Expanded Stochastic Gradient Markov Chain Monte Carlo (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.