English
Related papers

Related papers: Saddle-to-Saddle Dynamics in Diagonal Linear Netwo…

200 papers

Neural network optimization remains one of the most consequential yet poorly understood challenges in modern AI research, where improvements in training algorithms can lead to enhanced feature learning in foundation models,…

Machine Learning · Computer Science 2025-12-23 Ansh Nagwekar

We analyze the variance of stochastic gradients along negative curvature directions in certain non-convex machine learning models and show that stochastic gradients exhibit a strong component along these directions. Furthermore, we show…

Machine Learning · Computer Science 2018-09-18 Hadi Daneshmand , Jonas Kohler , Aurelien Lucchi , Thomas Hofmann

We consider linear models for stochastic dynamics. To any such model can be associated a network (namely a directed graph) describing which degrees of freedom interact under the dynamics. We tackle the problem of learning such a network…

Statistics Theory · Mathematics 2011-03-01 José Bento , Morteza Ibrahimi , Andrea Montanari

Adversarial training is a principled approach for training robust neural networks. Despite of tremendous successes in practice, its theoretical properties still remain largely unexplored. In this paper, we provide new theoretical insights…

Machine Learning · Computer Science 2019-07-29 Yan Li , Ethan X. Fang , Huan Xu , Tuo Zhao

We propose a derivative-free saddle-search algorithm designed to locate transition states using only function evaluations. The algorithm employs a nested architecture consisting of an inner eigenvector search and an outer saddle-point…

Numerical Analysis · Mathematics 2026-01-07 Qiang Du , Baoming Shi , Lei Zhang , Xiangcheng Zheng

In this paper we propose a primal-dual dynamical approach to the minimization of a structured convex function consisting of a smooth term, a nonsmooth term, and the composition of another nonsmooth term with a linear continuous operator. In…

Optimization and Control · Mathematics 2020-08-03 Radu Ioan Bot , Ernö Robert Csetnek , Szilard Laszlo

Neural network training is commonly based on SGD. However, the understanding of SGD's ability to converge to good local minima, given the non-convex nature of loss functions and the intricate geometric characteristics of loss landscapes,…

A recent line of work has shown that an overparametrized neural network can perfectly fit the training data, an otherwise often intractable nonconvex optimization problem. For (fully-connected) shallow networks, in the best case scenario,…

Machine Learning · Computer Science 2019-10-30 Armin Eftekhari , ChaeHwan Song , Volkan Cevher

The implicit bias induced by the training of neural networks has become a topic of rigorous study. In the limit of gradient flow and gradient descent with appropriate step size, it has been shown that when one trains a deep linear network…

Machine Learning · Computer Science 2022-04-27 Thien Le , Stefanie Jegelka

A central challenge to many fields of science and engineering involves minimizing non-convex error functions over continuous, high dimensional spaces. Gradient descent or quasi-Newton methods are almost ubiquitously used to perform such…

Machine Learning · Computer Science 2014-06-11 Yann Dauphin , Razvan Pascanu , Caglar Gulcehre , Kyunghyun Cho , Surya Ganguli , Yoshua Bengio

We consider distributed optimization under communication constraints for training deep learning models. We propose a new algorithm, whose parameter updates rely on two forces: a regular gradient step, and a corrective direction dictated by…

Machine Learning · Computer Science 2022-04-29 Yunfei Teng , Wenbo Gao , Francois Chalus , Anna Choromanska , Donald Goldfarb , Adrian Weller

We derive the system of differential equations for the gradient flow characterizing the training process of linear in-context learning in full generality. Next, we explore the geometric structure of the gradient flows in two instances,…

Dynamical Systems · Mathematics 2024-12-24 Songtao Lu , Yingdong Lu , Tomasz Nowicki

Stochastic gradient descent (SGD) is a standard optimization method to minimize a training error with respect to network parameters in modern neural network learning. However, it typically suffers from proliferation of saddle points in the…

Machine Learning · Computer Science 2017-11-23 Haiping Huang , Taro Toyoizumi

This paper establishes risk convergence and asymptotic weight matrix alignment --- a form of implicit regularization --- of gradient flow and gradient descent when applied to deep linear networks on linearly separable data. In more detail,…

Machine Learning · Computer Science 2019-02-26 Ziwei Ji , Matus Telgarsky

Learning rates in stochastic neural network training are currently determined a priori to training, using expensive manual or automated iterative tuning. This study proposes gradient-only line searches to resolve the learning rate for…

Machine Learning · Statistics 2020-01-16 Dominic Kafka , Daniel N. Wilke

The training dynamics of linear networks are well studied in two distinct setups: the lazy regime and balanced/active regime, depending on the initialization and width of the network. We provide a surprisingly simple unifying formula for…

Machine Learning · Computer Science 2024-10-31 Zhenfeng Tu , Santiago Aranguri , Arthur Jacot

We study continual learning on multiple linear classification tasks by sequentially running gradient descent (GD) for a fixed budget of iterations per task. When all tasks are jointly linearly separable and are presented in a cyclic/random…

Machine Learning · Computer Science 2025-04-29 Hyunji Jung , Hanseul Cho , Chulhee Yun

This paper proposes a certificate, rooted in observability, for asymptotic convergence of saddle flow dynamics of convex-concave functions to a saddle point. This observable certificate directly bridges the gap between the invariant set and…

Optimization and Control · Mathematics 2020-10-01 Pengcheng You , Enrique Mallada

We study the convergence dynamics of Gradient Descent (GD) in a minimal binary classification setting, consisting of a two-neuron ReLU network and two training instances. We prove that even under these strong simplifying assumptions, while…

Machine Learning · Computer Science 2026-03-03 Guy Smorodinsky , Sveta Gimpleson , Itay Safran

Many real world systems are at risk of undergoing critical transitions, leading to sudden qualitative and sometimes irreversible regime shifts. The development of early warning signals is recognized as a major challenge. Recent progress…

Adaptation and Self-Organizing Systems · Physics 2017-02-28 Christian Kuehn , Gerd Zschaler , Thilo Gross