On the Learning Dynamics of Two-layer Linear Networks with Label Noise SGD

• URL: http://arxiv.org/abs/2603.10397v1
• Date: Wed, 11 Mar 2026 04:26:54 GMT
• Title: On the Learning Dynamics of Two-layer Linear Networks with Label Noise SGD
• Abstract summary: We study the learning dynamics of gradient descent with label noise.<n>We highlight the critical role of label noise in driving the transition from the lazy to the rich regime.<n>We extend these insights to Sharpness-Aware Minimization (SAM), showing that the principles governing label noise SGD also apply to broader optimization algorithms.
• Score: 93.70725920710208
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: One crucial factor behind the success of deep learning lies in the implicit bias induced by noise inherent in gradient-based training algorithms. Motivated by empirical observations that training with noisy labels improves model generalization, we delve into the underlying mechanisms behind stochastic gradient descent (SGD) with label noise. Focusing on a two-layer over-parameterized linear network, we analyze the learning dynamics of label noise SGD, unveiling a two-phase learning behavior. In \emph{Phase I}, the magnitudes of model weights progressively diminish, and the model escapes the lazy regime; enters the rich regime. In \emph{Phase II}, the alignment between model weights and the ground-truth interpolator increases, and the model eventually converges. Our analysis highlights the critical role of label noise in driving the transition from the lazy to the rich regime and minimally explains its empirical success. Furthermore, we extend these insights to Sharpness-Aware Minimization (SAM), showing that the principles governing label noise SGD also apply to broader optimization algorithms. Extensive experiments, conducted under both synthetic and real-world setups, strongly support our theory. Our code is released at https://github.com/a-usually/Label-Noise-SGD.

Related papers

• Stochastic Gradient Descent in the Saddle-to-Saddle Regime of Deep Linear Networks [74.46751089984072]
  gradient descent (SGD) noise on descent in deep linear networks is poorly understood.<n>We investigate the dynamics of SGD during training of DLNs in the saddle-to-saddle regime.<n>Our results establish that SGD noise encodes information about the progression of feature learning but does not fundamentally alter the saddle-to-saddle dynamics.
  arXiv  Detail & Related papers  (2026-04-07T18:43:08Z)
• How Does Label Noise Gradient Descent Improve Generalization in the Low SNR Regime? [78.0226274470175]
  We investigate whether introducing label noise to the gradient updates can enhance the test performance of neural network (NN)<n>We prove that adding label noise during training suppresses noise memorization, preventing it from dominating the learning process.<n>In contrast, we show that NN trained with standard GD tends to overfit to noise in the same low SNR setting.
  arXiv  Detail & Related papers  (2025-10-20T13:28:13Z)
• Robust OOD Graph Learning via Mean Constraints and Noise Reduction [4.234842954797706]
  Graph Out-of-Distribution (OOD) classification often suffers from sharp performance drops, particularly under category imbalance and structural noise.<n>This work tackles two pressing challenges in this context: (1) the underperformance of minority classes due to skewed label distributions, and (2) their heightened sensitivity to structural noise in graph data.
  arXiv  Detail & Related papers  (2025-06-24T03:25:33Z)
• Stochastic Resetting Mitigates Latent Gradient Bias of SGD from Label Noise [2.048226951354646]
  We show that resetting from a checkpoint can significantly improve generalization performance when training deep neural networks (DNNs) with noisy labels.<n>In the presence of noisy labels, DNNs initially learn the general patterns of the data but then gradually memorize the corrupted data, leading to overfitting.<n>By deconstructing the dynamics of gradient descent (SGD), we identify the behavior of a latent gradient bias induced by noisy labels, which harms generalization.
  arXiv  Detail & Related papers  (2024-06-01T10:45:41Z)
• Doubly Stochastic Models: Learning with Unbiased Label Noises and Inference Stability [85.1044381834036]
  We investigate the implicit regularization effects of label noises under mini-batch sampling settings of gradient descent. We find such implicit regularizer would favor some convergence points that could stabilize model outputs against perturbation of parameters. Our work doesn't assume SGD as an Ornstein-Uhlenbeck like process and achieve a more general result with convergence of approximation proved.
  arXiv  Detail & Related papers  (2023-04-01T14:09:07Z)
• Dynamics-Aware Loss for Learning with Label Noise [73.75129479936302]
  Label noise poses a serious threat to deep neural networks (DNNs) We propose a dynamics-aware loss (DAL) to solve this problem. Both the detailed theoretical analyses and extensive experimental results demonstrate the superiority of our method.
  arXiv  Detail & Related papers  (2023-03-21T03:05:21Z)
• Latent Class-Conditional Noise Model [54.56899309997246]
  We introduce a Latent Class-Conditional Noise model (LCCN) to parameterize the noise transition under a Bayesian framework. We then deduce a dynamic label regression method for LCCN, whose Gibbs sampler allows us efficiently infer the latent true labels. Our approach safeguards the stable update of the noise transition, which avoids previous arbitrarily tuning from a mini-batch of samples.
  arXiv  Detail & Related papers  (2023-02-19T15:24:37Z)
• 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)
• Learning to Rectify for Robust Learning with Noisy Labels [25.149277009932423]
  We propose warped probabilistic inference (WarPI) to achieve adaptively rectifying the training procedure for the classification network. We evaluate WarPI on four benchmarks of robust learning with noisy labels and achieve the new state-of-the-art under variant noise types.
  arXiv  Detail & Related papers  (2021-11-08T02:25:50Z)

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.