AdaGrad-Diff: A New Version of the Adaptive Gradient Algorithm
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914328472453120 |
|---|---|
| author | Bojovic, Matia Salzo, Saverio Pontil, Massimiliano |
| author_facet | Bojovic, Matia Salzo, Saverio Pontil, Massimiliano |
| contents | Vanilla gradient methods are often highly sensitive to the choice of stepsize, which typically requires manual tuning. Adaptive methods alleviate this issue and have therefore become widely used. Among them, AdaGrad has been particularly influential. In this paper, we propose an AdaGrad-style adaptive method in which the adaptation is driven by the cumulative squared norms of successive gradient differences rather than gradient norms themselves. The key idea is that when gradients vary little across iterations, the stepsize is not unnecessarily reduced, while significant gradient fluctuations, reflecting curvature or instability, lead to automatic stepsize damping. Numerical experiments demonstrate that the proposed method is more robust than AdaGrad in several practically relevant settings. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_13112 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | AdaGrad-Diff: A New Version of the Adaptive Gradient Algorithm Bojovic, Matia Salzo, Saverio Pontil, Massimiliano Machine Learning Optimization and Control Vanilla gradient methods are often highly sensitive to the choice of stepsize, which typically requires manual tuning. Adaptive methods alleviate this issue and have therefore become widely used. Among them, AdaGrad has been particularly influential. In this paper, we propose an AdaGrad-style adaptive method in which the adaptation is driven by the cumulative squared norms of successive gradient differences rather than gradient norms themselves. The key idea is that when gradients vary little across iterations, the stepsize is not unnecessarily reduced, while significant gradient fluctuations, reflecting curvature or instability, lead to automatic stepsize damping. Numerical experiments demonstrate that the proposed method is more robust than AdaGrad in several practically relevant settings. |
| title | AdaGrad-Diff: A New Version of the Adaptive Gradient Algorithm |
| topic | Machine Learning Optimization and Control |
| url | https://arxiv.org/abs/2602.13112 |