Fast Last-Iterate Convergence of SGD in the Smooth Interpolation Regime
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913961800105984 |
|---|---|
| author | Attia, Amit Schliserman, Matan Sherman, Uri Koren, Tomer |
| author_facet | Attia, Amit Schliserman, Matan Sherman, Uri Koren, Tomer |
| contents | We study population convergence guarantees of stochastic gradient descent (SGD) for smooth convex objectives in the interpolation regime, where the noise at optimum is zero or near zero. The behavior of the last iterate of SGD in this setting -- particularly with large (constant) stepsizes -- has received growing attention in recent years due to implications for the training of over-parameterized models, as well as to analyzing forgetting in continual learning and to understanding the convergence of the randomized Kaczmarz method for solving linear systems. We establish that after $T$ steps of SGD on $β$-smooth convex loss functions with stepsize $0 < η< 2/β$, the last iterate exhibits expected excess risk $\widetilde{O}(\frac{1}{η(2-βη) T^{1-βη/2}} + \fracη{(2-βη)^2} T^{βη/2} σ_\star^2)$, where $σ_\star^2$ denotes the variance of the stochastic gradients at the optimum. In particular, for a well-tuned stepsize we obtain a near optimal $\widetilde{O}(1/T + σ_\star/\sqrt{T})$ rate for the last iterate, extending the results of Varre et al. (2021) beyond least squares regression; and when $σ_\star=0$ we obtain a rate of $\smash{O(1/\sqrt T)}$ with $η=1/β$, improving upon the best-known $\smash{O(T^{-1/4})}$ rate recently established by Evron et al. (2025) in the special case of realizable linear regression. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_11274 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Fast Last-Iterate Convergence of SGD in the Smooth Interpolation Regime Attia, Amit Schliserman, Matan Sherman, Uri Koren, Tomer Machine Learning Optimization and Control We study population convergence guarantees of stochastic gradient descent (SGD) for smooth convex objectives in the interpolation regime, where the noise at optimum is zero or near zero. The behavior of the last iterate of SGD in this setting -- particularly with large (constant) stepsizes -- has received growing attention in recent years due to implications for the training of over-parameterized models, as well as to analyzing forgetting in continual learning and to understanding the convergence of the randomized Kaczmarz method for solving linear systems. We establish that after $T$ steps of SGD on $β$-smooth convex loss functions with stepsize $0 < η< 2/β$, the last iterate exhibits expected excess risk $\widetilde{O}(\frac{1}{η(2-βη) T^{1-βη/2}} + \fracη{(2-βη)^2} T^{βη/2} σ_\star^2)$, where $σ_\star^2$ denotes the variance of the stochastic gradients at the optimum. In particular, for a well-tuned stepsize we obtain a near optimal $\widetilde{O}(1/T + σ_\star/\sqrt{T})$ rate for the last iterate, extending the results of Varre et al. (2021) beyond least squares regression; and when $σ_\star=0$ we obtain a rate of $\smash{O(1/\sqrt T)}$ with $η=1/β$, improving upon the best-known $\smash{O(T^{-1/4})}$ rate recently established by Evron et al. (2025) in the special case of realizable linear regression. |
| title | Fast Last-Iterate Convergence of SGD in the Smooth Interpolation Regime |
| topic | Machine Learning Optimization and Control |
| url | https://arxiv.org/abs/2507.11274 |