arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2401.04067cs.LGmath.STstat.TH
收藏

凸SGD:无需提前停止的泛化

Convex SGD: Generalization Without Early Stopping

UCLouvain(卢汶大学);Boston University(波士顿大学)

Julien Hendrickx, Alex Olshevsky

AI总结:

该研究针对紧集上光滑凸函数的随机梯度下降泛化问题,首次推导了迭代次数与数据集规模任意增长时趋于零的泛化误差界,证明无需强凸性即可实现良好泛化。

摘要:

我们研究紧集上光滑凸函数的随机梯度下降泛化误差。首次给出当迭代次数$T$和数据集规模$n$以任意速率趋于无穷时趋于零的泛化误差界;在步长$α_t = 1/\sqrt{t}$下,误差界为$\tilde{O}(1/\sqrt{T} + 1/\sqrt{n})$。特别地,随机梯度下降实现良好泛化不需要强凸性。

英文摘要:

We consider the generalization error associated with stochastic gradient descent on a smooth convex function over a compact set. We show the first bound on the generalization error that vanishes when the number of iterations $T$ and the dataset size $n$ go to zero at arbitrary rates; our bound scales as $\tilde{O}(1/\sqrt{T} + 1/\sqrt{n})$ with step-size $α_t = 1/\sqrt{t}$. In particular, strong convexity is not needed for stochastic gradient descent to generalize well.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑