中文
相关论文

相关论文: Catastrophic Fisher Explosion: Early Phase Fisher …

200 篇论文

Deep Reinforcement Learning (DRL) systems often tend to overfit to early experiences, a phenomenon known as the primacy bias (PB). This bias can severely hinder learning efficiency and final performance, particularly in complex…

机器学习 · 计算机科学 2025-02-04 Massimiliano Falzari , Matthia Sabatelli

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

Prior research shows that differences in the early period of neural network training significantly impact the performance of in-distribution (ID) data of tasks. Yet, the implications of early learning dynamics on out-of-distribution (OOD)…

机器学习 · 计算机科学 2025-01-07 Chen Cecilia Liu , Iryna Gurevych

Normalization methods play an important role in enhancing the performance of deep learning while their theoretical understandings have been limited. To theoretically elucidate the effectiveness of normalization, we quantify the geometry of…

机器学习 · 统计学 2019-10-29 Ryo Karakida , Shotaro Akaho , Shun-ichi Amari

A number of competing hypotheses have been proposed to explain why small-batch Stochastic Gradient Descent (SGD)leads to improved generalization over the full-batch regime, with recent work crediting the implicit regularization of various…

机器学习 · 计算机科学 2022-11-30 Zachary Novack , Simran Kaur , Tanya Marwah , Saurabh Garg , Zachary C. Lipton

The Fisher information matrix (FIM) is a fundamental quantity to represent the characteristics of a stochastic model, including deep neural networks (DNNs). The present study reveals novel statistics of FIM that are universal among a wide…

机器学习 · 统计学 2019-10-10 Ryo Karakida , Shotaro Akaho , Shun-ichi Amari

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

In the realm of deep learning, the Fisher information matrix (FIM) gives novel insights and useful tools to characterize the loss landscape, perform second-order optimization, and build geometric learning theories. The exact FIM is either…

机器学习 · 计算机科学 2021-10-29 Alexander Soen , Ke Sun

For infinitesimal learning rates, stochastic gradient descent (SGD) follows the path of gradient flow on the full batch loss function. However moderately large learning rates can achieve higher test accuracies, and this generalization…

机器学习 · 计算机科学 2021-01-29 Samuel L. Smith , Benoit Dherin , David G. T. Barrett , Soham De

Randomly initialized neural networks induce a prior over functions, but the predictor used in practice is produced only after training. We ask how much of this initial bias survives the training pipeline. To make the question measurable, we…

机器学习 · 计算机科学 2026-05-29 Mohua Das , Pierfrancesco Beneventano , Shibshankar Dey , Gareth H. McKinkey , Tomaso Poggio

The early phase of training of deep neural networks is critical for their final performance. In this work, we study how the hyperparameters of stochastic gradient descent (SGD) used in the early phase of training affect the rest of the…

We derive a differential equation that governs the evolution of the generalization gap when a deep network is trained by gradient descent. This differential equation is controlled by two quantities, a contraction factor that brings together…

机器学习 · 计算机科学 2025-10-14 Rubing Yang , Pratik Chaudhari

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

Machine learning models trained with \emph{stochastic} gradient descent (SGD) can generalize better than those trained with deterministic gradient descent (GD). In this work, we study SGD's impact on generalization through the lens of the…

机器学习 · 计算机科学 2025-12-09 Hongjian Lan , Yucong Liu , Florian Schäfer

This paper introduces a new stochastic optimization method based on the regularized Fisher information matrix (FIM), named SOFIM, which can efficiently utilize the FIM to approximate the Hessian matrix for finding Newton's gradient update…

机器学习 · 计算机科学 2024-05-02 Mrinmay Sen , A. K. Qin , Gayathri C , Raghu Kishore N , Yen-Wei Chen , Balasubramanian Raman

Several works have aimed to explain why overparameterized neural networks generalize well when trained by Stochastic Gradient Descent (SGD). The consensus explanation that has emerged credits the randomized nature of SGD for the bias of the…

机器学习 · 计算机科学 2021-02-24 Shengchao Liu , Dimitris Papailiopoulos , Dimitris Achlioptas

The mechanisms by which certain training interventions, such as increasing learning rates and applying batch normalization, improve the generalization of deep networks remains a mystery. Prior works have speculated that "flatter" solutions…

机器学习 · 计算机科学 2023-05-25 Simran Kaur , Jeremy Cohen , Zachary C. Lipton

The Fisher information matrix (FIM) is fundamental to understanding the trainability of deep neural nets (DNN), since it describes the parameter space's local metric. We investigate the spectral distribution of the conditional FIM, which is…

机器学习 · 统计学 2021-03-31 Tomohiro Hayase , Ryo Karakida

In this paper, we first present an explanation regarding the common occurrence of spikes in the training loss when neural networks are trained with stochastic gradient descent (SGD). We provide evidence that the spikes in the training loss…

机器学习 · 计算机科学 2024-06-07 Libin Zhu , Chaoyue Liu , Adityanarayanan Radhakrishnan , Mikhail Belkin

The use of low-bit quantization has emerged as an indispensable technique for enabling the efficient training of large-scale models. Despite its widespread empirical success, a rigorous theoretical understanding of its impact on learning…

机器学习 · 统计学 2026-02-17 Dechen Zhang , Junwei Su , Difan Zou
‹ 上一页 1 2 3 10 下一页 ›