中文
相关论文

相关论文: Shape Matters: Understanding the Implicit Bias of …

200 篇论文

In overparametrized models, the noise in stochastic gradient descent (SGD) implicitly regularizes the optimization trajectory and determines which local minimum SGD converges to. Motivated by empirical studies that demonstrate that training…

机器学习 · 计算机科学 2021-12-07 Alex Damian , Tengyu Ma , Jason D. Lee

Random label noises (or observational noises) widely exist in practical machine learning settings. While previous studies primarily focus on the affects of label noises to the performance of learning, our work intends to investigate the…

机器学习 · 计算机科学 2023-04-04 Haoyi Xiong , Xuhong Li , Boyang Yu , Zhanxing Zhu , Dongrui Wu , Dejing Dou

Stochastic gradient descent (SGD) forms the core optimization method for deep neural networks. While some theoretical progress has been made, it still remains unclear why SGD leads the learning dynamics in overparameterized networks to…

机器学习 · 计算机科学 2019-10-30 Mingwei Wei , David J Schwab

Understanding the implicit bias of Stochastic Gradient Descent (SGD) is one of the key challenges in deep learning, especially for overparametrized models, where the local minimizers of the loss function $L$ can form a manifold.…

机器学习 · 计算机科学 2022-07-29 Zhiyuan Li , Tianhao Wang , Sanjeev Arora

In this paper, we provide a theoretical study of noise geometry for minibatch stochastic gradient descent (SGD), a phenomenon where noise aligns favorably with the geometry of local landscape. We propose two metrics, derived from analyzing…

机器学习 · 计算机科学 2024-02-02 Mingze Wang , Lei Wu

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…

机器学习 · 计算机科学 2026-03-12 Tongcheng Zhang , Zhanpeng Zhou , Mingze Wang , Andi Han , Wei Huang , Taiji Suzuki , Junchi Yan

We develop generalization error bounds for stochastic gradient descent (SGD) with label noise in non-convex settings under uniform dissipativity and smoothness conditions. Under a suitable choice of semimetric, we establish a contraction in…

机器学习 · 统计学 2023-11-02 Jung Eun Huh , Patrick Rebeschini

The gradient noise of SGD is considered to play a central role in the observed strong generalization abilities of deep learning. While past studies confirm that the magnitude and the covariance structure of gradient noise are critical for…

机器学习 · 计算机科学 2020-06-22 Jingfeng Wu , Wenqing Hu , Haoyi Xiong , Jun Huan , Vladimir Braverman , Zhanxing Zhu

Understanding the implicit bias of training algorithms is of crucial importance in order to explain the success of overparametrised neural networks. In this paper, we study the role of the label noise in the training dynamics of a…

机器学习 · 统计学 2022-06-22 Loucas Pillaud-Vivien , Julien Reygner , Nicolas Flammarion

It is often observed that stochastic gradient descent (SGD) and its variants implicitly select a solution with good generalization performance; such implicit bias is often characterized in terms of the sharpness of the minima. Kleinberg et…

机器学习 · 统计学 2024-05-28 Atsushi Nitanda , Ryuhei Kikuchi , Shugo Maeda , Denny Wu

Training Deep Neural Networks (DNNs) with small batches using Stochastic Gradient Descent (SGD) yields superior test performance compared to larger batches. The specific noise structure inherent to SGD is known to be responsible for this…

机器学习 · 统计学 2024-02-14 Tom Sander , Maxime Sylvestre , Alain Durmus

Stochastic Gradient Descent (SGD) is the workhorse algorithm of deep learning technology. At each step of the training phase, a mini batch of samples is drawn from the training dataset and the weights of the neural network are adjusted…

无序系统与神经网络 · 物理学 2022-09-07 Francesca Mignacco , Pierfrancesco Urbani

Previous work has examined the ability of larger capacity neural networks to generalize better than smaller ones, even without explicit regularizers, by analyzing gradient based algorithms such as GD and SGD. The presence of noise and its…

机器学习 · 计算机科学 2020-05-27 Arushi Gupta

Deep neural networks with remarkably strong generalization performances are usually over-parameterized. Despite explicit regularization strategies are used for practitioners to avoid over-fitting, the impacts are often small. Some…

计算与语言 · 计算机科学 2018-11-05 Deren Lei , Zichen Sun , Yijun Xiao , William Yang Wang

Stochastic gradient descent (SGD) is one of the most popular algorithms in modern machine learning. The noise encountered in these applications is different from that in many theoretical analyses of stochastic gradient algorithms. In this…

机器学习 · 统计学 2021-09-16 Stephan Wojtowytsch

The classical statistical learning theory implies that fitting too many parameters leads to overfitting and poor performance. That modern deep neural networks generalize well despite a large number of parameters contradicts this finding and…

机器学习 · 统计学 2022-10-18 Masaaki Imaizumi , Johannes Schmidt-Hieber

The phenomenon that stochastic gradient descent (SGD) favors flat minima has played a critical role in understanding the implicit regularization of SGD. In this paper, we provide an explanation of this striking phenomenon by relating the…

机器学习 · 统计学 2022-10-18 Lei Wu , Mingze Wang , Weijie Su

Stochastic Gradient Descent (SGD) has become a cornerstone of neural network optimization due to its computational efficiency and generalization capabilities. However, the gradient noise introduced by SGD is often assumed to be uncorrelated…

机器学习 · 计算机科学 2025-12-23 Marcel Kühn , Bernd Rosenow

In this work, we reveal a strong implicit bias of stochastic gradient descent (SGD) that drives overly expressive networks to much simpler subnetworks, thereby dramatically reducing the number of independent parameters, and improving…

机器学习 · 计算机科学 2024-05-30 Feng Chen , Daniel Kunin , Atsushi Yamamura , Surya Ganguli

We interpret the variational inference of the Stochastic Gradient Descent (SGD) as minimizing a new potential function named the \textit{quasi-potential}. We analytically construct the quasi-potential function in the case when the loss…

机器学习 · 计算机科学 2019-01-21 Wenqing Hu , Zhanxing Zhu , Haoyi Xiong , Jun Huan
‹ 上一页 1 2 3 10 下一页 ›