Global Dynamics of Heavy-Tailed SGDs in Nonconvex Loss Landscape: Characterization and Control

• URL: http://arxiv.org/abs/2510.20905v1
• Date: Thu, 23 Oct 2025 18:01:29 GMT
• Title: Global Dynamics of Heavy-Tailed SGDs in Nonconvex Loss Landscape: Characterization and Control
• Authors: Xingyu Wang, Chang-Han Rhee,
• Abstract summary: gradient descent (SGD) and its variants enable modern artificial intelligence.<n>It is widely believed that SGD has a curious ability to avoid sharp local minima in the loss landscape.<n>We reveal a fascinating phenomenon in deep learning: by injecting and then truncating heavy-tailed noises during the training phase, SGD can almost completely avoid sharp minima.
• Score: 7.665296591586615
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Stochastic gradient descent (SGD) and its variants enable modern artificial intelligence. However, theoretical understanding lags far behind their empirical success. It is widely believed that SGD has a curious ability to avoid sharp local minima in the loss landscape, which are associated with poor generalization. To unravel this mystery and further enhance such capability of SGDs, it is imperative to go beyond the traditional local convergence analysis and obtain a comprehensive understanding of SGDs' global dynamics. In this paper, we develop a set of technical machinery based on the recent large deviations and metastability analysis in Wang and Rhee (2023) and obtain sharp characterization of the global dynamics of heavy-tailed SGDs. In particular, we reveal a fascinating phenomenon in deep learning: by injecting and then truncating heavy-tailed noises during the training phase, SGD can almost completely avoid sharp minima and achieve better generalization performance for the test data. Simulation and deep learning experiments confirm our theoretical prediction that heavy-tailed SGD with gradient clipping finds local minima with a more flat geometry and achieves better generalization performance.

Related papers

• On the Convergence of (Stochastic) Gradient Descent for Kolmogorov--Arnold Networks [56.78271181959529]
  Kolmogorov--Arnold Networks (KANs) have gained significant attention in the deep learning community. Empirical investigations demonstrate that KANs optimized via gradient descent (SGD) are capable of achieving near-zero training loss.
  arXiv  Detail & Related papers  (2024-10-10T15:34:10Z)
• The Optimality of (Accelerated) SGD for High-Dimensional Quadratic Optimization [4.7256945641654164]
  gradient descent (SGD) is a widely used algorithm in machine learning, particularly for neural network training. Recent studies on SGD for canonical quadratic optimization or linear regression show it attains well generalization under suitable high-dimensional settings. This paper investigates SGD with two essential components in practice: exponentially decaying step size schedule and momentum.
  arXiv  Detail & Related papers  (2024-09-15T14:20:03Z)
• SGD with Large Step Sizes Learns Sparse Features [22.959258640051342]
  We showcase important features of the dynamics of the Gradient Descent (SGD) in the training of neural networks. We show that the longer large step sizes keep SGD high in the loss landscape, the better the implicit regularization can operate and find sparse representations.
  arXiv  Detail & Related papers  (2022-10-11T11:00:04Z)
• Stability and Generalization Analysis of Gradient Methods for Shallow Neural Networks [59.142826407441106]
  We study the generalization behavior of shallow neural networks (SNNs) by leveraging the concept of algorithmic stability. We consider gradient descent (GD) and gradient descent (SGD) to train SNNs, for both of which we develop consistent excess bounds.
  arXiv  Detail & Related papers  (2022-09-19T18:48:00Z)
• Low-Precision Stochastic Gradient Langevin Dynamics [70.69923368584588]
  We provide the first study of low-precision Gradient Langevin Dynamics, showing that its costs can be significantly reduced without sacrificing performance. We develop a new quantization function for SGLD that preserves the variance in each update step. We demonstrate that low-precision SGLD achieves comparable performance to full-precision SGLD with only 8 bits on a variety of deep learning tasks.
  arXiv  Detail & Related papers  (2022-06-20T17:25:41Z)
• Implicit Regularization or Implicit Conditioning? Exact Risk Trajectories of SGD in High Dimensions [26.782342518986503]
  gradient descent (SGD) is a pillar of modern machine learning, serving as the go-to optimization algorithm for a diverse array of problems. We show how to adapt the HSGD formalism to include streaming SGD, which allows us to produce an exact prediction for the excess risk of multi-pass SGD relative to that of streaming SGD.
  arXiv  Detail & Related papers  (2022-06-15T02:32:26Z)
• Stochastic Training is Not Necessary for Generalization [57.04880404584737]
  It is widely believed that the implicit regularization of gradient descent (SGD) is fundamental to the impressive generalization behavior we observe in neural networks. In this work, we demonstrate that non-stochastic full-batch training can achieve strong performance on CIFAR-10 that is on-par with SGD.
  arXiv  Detail & Related papers  (2021-09-29T00:50:00Z)
• Eliminating Sharp Minima from SGD with Truncated Heavy-tailed Noise [39.27123042800951]
  Evidence of heavy-tailed gradient noise was reported in many deep learning tasks. We show that truncated SGD can effectively eliminate sharp local minima entirely from its training trajectory.
  arXiv  Detail & Related papers  (2021-02-08T16:03:49Z)
• Direction Matters: On the Implicit Bias of Stochastic Gradient Descent with Moderate Learning Rate [105.62979485062756]
  This paper attempts to characterize the particular regularization effect of SGD in the moderate learning rate regime. We show that SGD converges along the large eigenvalue directions of the data matrix, while GD goes after the small eigenvalue directions.
  arXiv  Detail & Related papers  (2020-11-04T21:07:52Z)
• On the Generalization of Stochastic Gradient Descent with Momentum [84.54924994010703]
  momentum-based accelerated variants of gradient descent (SGD) are widely used when training machine learning models. We first show that there exists a convex loss function for which the stability gap for multiple epochs of SGD with standard heavy-ball momentum (SGDM) becomes unbounded. For smooth Lipschitz loss functions, we analyze a modified momentum-based update rule, i.e., SGD with early momentum (SGDEM) under a broad range of step-sizes.
  arXiv  Detail & Related papers  (2018-09-12T17:02:08Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.