Global ETD Search

Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.

Results

Showing 1 to 12 of 12 for “"Stochastic Gradient Langevin Dynamics"”.

  1. SCALABLE BAYESIAN INFERENCE FOR LARGE CROSSED MIXED EFFECTS MODELS

    … sampling algorithms. We propose a class of stochastic gradient MCMC algorithms for such models. Our first contribution is to devise novel algorithms for the crossed mixed effects models for continuous response variables. The first algorithm is developed for the balanced design. The second …

    nus Repository record for SCALABLE BAYESIAN INFERENCE FOR LARGE CROSSED MIXED EFFECTS MODELS (opens in a new tab) Download file for SCALABLE BAYESIAN INFERENCE FOR LARGE CROSSED MIXED EFFECTS MODELS (opens in a new tab)

  2. Batch Normalization Preconditioning for Neural Network Training

    … normalization by conditioning the parameter gradients directly during training. This is designed to improve the Hessian matrix of the loss function and hence convergence during training. One benefit is that BNP is not constrained on the mini-batch size and works in the online learning …

    uky-thes Repository record for Batch Normalization Preconditioning for Neural Network Training (opens in a new tab)

  3. Are Particle-Based Methods the Future of Sampling in Joint Energy Models? A Deep Dive into SVGD and SGLD

    … the integration of Stein Variational Gradient Descent (SVGD) with Joint Energy Models (JEMs), comparing its performance to Stochastic Gradient Langevin Dynamics (SGLD). We incorporated a generative loss term with an entropy component to enhance diversity and a smoothing factor to …

    vt Repository record for Are Particle-Based Methods the Future of Sampling in Joint Energy Models? A Deep Dive into SVGD and SGLD (opens in a new tab)

  4. INVERSE PROBLEMS IN IMAGING: FROM REGULARIZATION METHODS TO SELF-SUPERVISED LEARNING METHODS

    … reparameterize the image and apply under-damped Langevin dynamics as the sampling method. For linear inverse problems, we present a self-supervised learning approach that is based on the neuralization of a Bayesian estimator of the problem, The key ingredient of the neuralized estimator is an …

    nus Repository record for INVERSE PROBLEMS IN IMAGING: FROM REGULARIZATION METHODS TO SELF-SUPERVISED LEARNING METHODS (opens in a new tab) Download file for INVERSE PROBLEMS IN IMAGING: FROM REGULARIZATION METHODS TO SELF-SUPERVISED LEARNING METHODS (opens in a new tab)

  5. Monte-Carlo neural rendering

    … regions with higher reconstruction error. Using Langevin Monte Carlo, sampling probabilities adapt dynamically during training, improving both convergence speed and final rendering quality. Second, we reformulate 3DGS as a Markov Chain Monte Carlo process, interpreting Gaussians as probabilistic …

    ubc Repository record for Monte-Carlo neural rendering (opens in a new tab) Download PDF for Monte-Carlo neural rendering (opens in a new tab)

  6. Enhancing the reliability of general-purpose algorithms for approximate Bayesian inference

    … posterior approximation algorithms such as stochastic gradient Langevin dynamics (SGLD) and variational inference offer both flexibility and computational efficiency. While essential for making reliable approximation inference, robust validation of approximation quality and accurate …

    bu Repository record for Enhancing the reliability of general-purpose algorithms for approximate Bayesian inference (opens in a new tab)

  7. Information-Theoretic Measures of Generalization in Machine Learning

    … in studying the generalization error of Stochastic Gradient Langevin Dynamics (SGLD). Our numerical analyses show that the obtained bounds are the first non-vacuous generalization guarantees for SGLD and its full-batch counterpart for modern deep-learning datasets and architectures. …

    toronto-retro Repository record for Information-Theoretic Measures of Generalization in Machine Learning (opens in a new tab)

  8. Advances in distributed Bayesian inference and graph neural networks

    … the same front, we also propose an extension of stochastic gradient Langevin dynamics for federated data, which are inherently distributed in a non-IID manner and cannot be centralized due to privacy constraints. Second, we develop a methodology for meta-analysis which allows the combination of …

    aalto Repository record for Advances in distributed Bayesian inference and graph neural networks (opens in a new tab)

  9. Solving sampling and optimization problems via Tamed Langevin MCMC algorithms in the presence of super-linearities

    … distributions with super-linearly growing log-gradient and the applications to the related optimization problems. This challenge is tackled in different scenarios using explicit numerical schemes which have their foundations on the concept of taming, which is a technique deployed to remedy the …

    greece Repository record for Solving sampling and optimization problems via Tamed Langevin MCMC algorithms in the presence of super-linearities (opens in a new tab)

  10. INTERPRETABLE DEEP LEARNING AND GAUSSIAN PROCESS MODELS FOR DISCRETE CHOICE

    … approximate Bayesian inference methods such as Stochastic Gradient Langevin Dynamics (SGLD). The model balances predictive performance with interpretability and uncertainty quantification, addressing key limitations of standard deep learning in economic applications. It adapts to data …

    cornell Repository record for INTERPRETABLE DEEP LEARNING AND GAUSSIAN PROCESS MODELS FOR DISCRETE CHOICE (opens in a new tab)

  11. Topics in sparse Bayesian machine learning

    … can be further reduced by data sub-sampling when stochastic gradient Langevin dynamics are employed. The algorithm is an extension of the asynchronous Gibbs sampler of Johnson et al. (2013), but can be viewed from a statistical perspective as a form of Bayesian iterated sure independent screening …

    bu Repository record for Topics in sparse Bayesian machine learning (opens in a new tab)

  12. Noisy Iterative Learning Algorithms: Optimization, Generalization, and Privacy

    … performance of deep networks is limited. Gradient-based methods have been popular to perform optimization and train neural networks. One promising direction towards tackling the above challenges is the noisy, iterative algorithm which adds extra noise in updating the parameters. The noise …

    umn Repository record for Noisy Iterative Learning Algorithms: Optimization, Generalization, and Privacy (opens in a new tab)