mlsanityData to 7 Oct 2026

Non-convex Learning via Replica Exchange Stochastic Gradient MCMC

arXiv 2008.05367 ↗First posted 12 Aug 2020stat.ML cs.LG math.PRcited by 33

Not on the map. Why?

Abstract (arXiv)

Replica exchange Monte Carlo (reMC), also known as parallel tempering, is an important technique for accelerating the convergence of the conventional Markov Chain Monte Carlo (MCMC) algorithms. However, such a method requires the evaluation of the energy function based on the full dataset and is not scalable to big data. The na"ive implementation of reMC in mini-batch settings introduces large biases, which cannot be directly extended to the stochastic gradient MCMC (SGMCMC), the standard sampling method for simulating from deep neural networks (DNNs). In this paper, we propose an adaptive replica exchange SGMCMC (reSGMCMC) to automatically correct the bias and study the corresponding properties. The analysis implies an acceleration-accuracy trade-off in the numerical discretization of a Markov jump process in a stochastic environment. Empirically, we test the algorithm through extensive experiments on various setups and obtain the state-of-the-art results on CIFAR10, CIFAR100, and SVHN in both supervised learning and semi-supervised learning tasks.

Key facts. Cited by 33 papers in the map (data through 7 Oct 2026). First posted on arXiv 12 Aug 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 5; 2022: 4; 2023: 4; 2024: 6; 2025: 12; 2026: 2.

33 papers.

Sort:

By area:

  1. 2024Conformalized-DeepONet: A Distribution-Free Framework for Uncertainty Quantification in Deep Operator Networkscited by 28
  2. 2022Bayesian autoencoders for data-driven discovery of coordinates, governing equations and fundamental constantscited by 24
  3. 2021Accelerated replica exchange stochastic gradient Langevin diffusion enhanced Bayesian DeepONet for solving noisy parametric PDEscited by 22
  4. 2021On Convergence of Federated Averaging Langevin Dynamicscited by 14
  5. 2021Polygonal Unadjusted Langevin Algorithms: Creating stable and efficient adaptive algorithms for neural networkscited by 11
  6. 2022Scalable Bayesian Uncertainty Quantification for Neural Network Potentials: Promise and Pitfallscited by 11
  7. 2023Deep Operator Learning-based Surrogate Models with Uncertainty Quantification for Optimizing Internal Cooling Channel Rib Profilescited by 10
  8. 2024Accelerating Approximate Thompson Sampling with Underdamped Langevin Monte Carlocited by 10
  9. 2022Interacting Contour Stochastic Gradient Langevin Dynamicscited by 9
  10. 2024Gradient-based Discrete Sampling with Automatic Cyclical Schedulingcited by 9
  11. 2021UPANets: Learning from the Universal Pixel Attention Networkscited by 7
  12. 2022Non-reversible Parallel Tempering for Deep Posterior Approximationcited by 7
21 more
  1. 2024Constrained Exploration via Reflected Replica Exchange Stochastic Gradient Langevin Dynamicscited by 7
  2. 2025Accelerating Langevin Monte Carlo Sampling: A Large Deviations Analysiscited by 4
  3. 2024Hard-Attention Gates with Gradient Routing for Endoscopic Image Computingcited by 3
  4. 2025Morephy-Net: An Evolutionary Multi-objective Optimization for Replica-Exchange-based Physics-informed Neural Operator Learning Networkscited by 3
  5. 2025Active Learning with Bayesian Multi-Fidelity Laplace Neural Operators for Oscillatory Parametric PDEscited by 2
  6. 2025Parameter Expanded Stochastic Gradient Markov Chain Monte Carlocited by 2
  7. 2023Subsampling Error in Stochastic Gradient Langevin Diffusionscited by 1
  8. 2023B-LSTM-MIONet: Bayesian LSTM-based Neural Operators for Learning the Response of Complex Dynamical Systems to Length-Variant Multiple Input Functionscited by 1
  9. 2025Bayesian Computation in Deep Learningcited by 1
  10. 2025Enhancing Gradient-based Discrete Sampling via Parallel Temperingcited by 1
  11. 2025JaxSGMC: Modular stochastic gradient MCMC in JAXcited by 1
  12. 2025Continuous Policy and Value Iteration for Relaxed Stochastic Control Problems and Its Convergencecited by 1
  13. 2025One Period to Rule Them All: Identifying Critical Learning Periods in Deep Networkscited by 1
  14. 2021A New Framework for Variance-Reduced Hamiltonian Monte Carlocited by 0
  15. 2023The surrogate Gibbs-posterior of a corrected stochastic MALA: Towards uncertainty quantification for neural networkscited by 0
  16. 2024Feynman-Kac Operator Expectation Estimatorcited by 0
  17. 2025Exploring Non-Convex Discrete Energy Landscapes: An Efficient Langevin-Like Sampler with Replica Exchangecited by 0
  18. 2025BLADE: Bayesian Langevin Active Discovery with Replica Exchange for Identification of Complex Systemscited by 0
  19. 2025Flatness-Aware Stochastic Gradient Langevin Dynamicscited by 0
  20. 2026Mode Coverage in Normalizing Flow Boltzmann Generators via Log-Ratio Variationcited by 0
  21. 2026Explore Broadly, Reason Sharply: Push Small Models toward the Frontier via Samplingcited by 0