The Last Iterate of the Projected Stochastic Subgradient Method Pays a Logarithm, Even in Dimension One
We prove that the projected stochastic subgradient method with the standard anytime stepsizes $η_t = γ/\sqrt t$ can suffer expected last-iterate optimization error of order $(\log n)/\sqrt n$ already on the real line. Combined with the classical in-expectation upper bound, this establishes the sharp worst-case order of the projected stochastic subgradient method.
Comments and reviews
No comments yet.
API
curl /api/v1/records/arxiv/2610.30143
curl -X POST /api/v1/records/arxiv/2610.30143/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2610.30143/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"body": "…", "review": true}'