Revisiting AdaGrad in Stochastic Convex Optimization: Last Iterates, High Probability, and Lower Bounds PGJ37WRB1K
| URI | m2://res/PGJ37WRB1K |
|---|---|
| Kind | preprint |
| Creators | Weiming Ou, Xiao Wang |
| Date | 2026-09-26 |
| MSC | 49-XX, 90-XX, 93-XX |
| Subjects | arxiv math.OC, arxiv cs.LG |
| Scope | in |
Expressions
e1: version, preprint
| Language | en |
|---|---|
| Date | 2026-09-26 |
| Ids | arXiv 2609.32729 |
Items in this resource
The mentions of the inventory that name this resource. Each name links to the card of its inventory row.
| Item | Kind | Where | Source |
|---|---|---|---|
| 1HWMPR23HH | definition | definition | arxiv |
| Constant worst-case error under bounded variance | statement | theorem 3.8 | arxiv |
| A scalar logarithmic average-error lower bound | statement | theorem 3.10 | arxiv |
| No polynomial average-iterate rate with polylogarithmic confidence | statement | theorem 3.5 | arxiv |
| Noise-adaptive average-iterate upper bounds | statement | theorem 3.14 | arxiv |
| Arbitrarily slow last-iterate convergence | statement | theorem 3.1 | arxiv |
Catalogue records
- arxiv: 2609.32729 (work)
- arxiv_text: 2609.32729 (work)