paper-with-me

홈 › Papers

Understanding the unstable convergence of gradient descent

2022-04-03 · Kwangjun Ahn, Jingzhao Zhang, Suvrit Sra

Most existing analyses of (stochastic) gradient descent rely on the condition that for $L$-smooth costs, the step size is less than $2/L$. However, many works have observed that in machine learning applications step sizes often do not fulfill this condition, yet (stochastic) gradient descent still converges, albeit in an unstable manner. We investigate this unstable convergence phenomenon from first principles, and discuss key causes behind it. We also identify its main characteristics, and how they interrelate based on both theory and experiments, offering a principled view toward understanding the phenomenon.

📄 PDF Abstract BibTeX arXiv:2204.01050

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Exponential Convergence of (Stochastic) Gradient Descent for Separable Logistic Regression

2026-02-21 · Sacchit Kale, Piyushi Manupriya, Pierre Marion, Francis Bach 외 arxiv

Gradient descent and stochastic gradient descent are central to modern machine learning, yet their behavior under large step sizes remains theoretically unclear. Recent work suggests that acceleration often arises near t…

Survey Descent: A Multipoint Generalization of Gradient Descent for Nonsmooth Optimization

2021-11-30 · X. Y. Han, Adrian S. Lewis

For strongly convex objectives that are smooth, the classical theory of gradient descent ensures linear convergence relative to the number of gradient evaluations. An analogous nonsmooth theory is challenging. Even when …

The Limit Points of (Optimistic) Gradient Descent in Min-Max Optimization

2018-07-11 · NeurIPS 2018 12 · Constantinos Daskalakis, Ioannis Panageas

Motivated by applications in Optimization, Game Theory, and the training of Generative Adversarial Networks, the convergence properties of first order methods in min-max problems have received extensive study. It has bee…

Learning Unstable Dynamical Systems with Time-Weighted Logarithmic Loss

2020-07-10 · Kamil Nar, Yuan Xue, Andrew M. Dai

When training the parameters of a linear dynamical model, the gradient descent algorithm is likely to fail to converge if the squared-error loss is used as the training loss function. Restricting the parameter space to a…

Interaction Matters: A Note on Non-asymptotic Local Convergence of Generative Adversarial Networks

2018-02-16 · Tengyuan Liang, James Stokes

Motivated by the pursuit of a systematic computational and algorithmic understanding of Generative Adversarial Networks (GANs), we present a simple yet unified non-asymptotic local convergence theory for smooth two-playe…