论文标题
了解深度学习稳定边缘的梯度下降
Understanding Gradient Descent on Edge of Stability in Deep Learning
论文作者
论文摘要
Cohen等人的深度学习实验。 [2021]使用确定性梯度下降（GD）显示学习率（LR）和清晰度（即Hessian的最大特征值）的稳定边缘（EOS）阶段不再像传统优化一样行为。清晰度稳定在$ 2/$ LR的左右，并且在迭代中损失不断上下，但仍有整体下降趋势。当前的论文数学分析了EOS阶段中隐式正则化的新机制，在该机构中，由于非平滑损失景观而导致的GD更新沿最小损失的多种流动上的某些确定性流程发展。这与许多先前关于隐式偏差依靠无限更新或梯度中的噪声的结果相反。正式地，对于具有某些规律性条件的任何平滑函数$ l $，对于（1）标准化的GD，即具有不同的lr $η_t= \fracη{\ |的GD证明了此效果。 \ nabla l（x（t））\ |} $和损失$ l $; （2）具有常数LR和损失$ \ sqrt {l- \ min_x l（x）} $的GD。两者都可以证明进入稳定性的边缘，在歧管上的相关流量最小化$λ_{1}（\ nabla^2 l）$。一项实验研究证实了上述理论结果。
Deep learning experiments by Cohen et al. [2021] using deterministic Gradient Descent (GD) revealed an Edge of Stability (EoS) phase when learning rate (LR) and sharpness (i.e., the largest eigenvalue of Hessian) no longer behave as in traditional optimization. Sharpness stabilizes around $2/$LR and loss goes up and down across iterations, yet still with an overall downward trend. The current paper mathematically analyzes a new mechanism of implicit regularization in the EoS phase, whereby GD updates due to non-smooth loss landscape turn out to evolve along some deterministic flow on the manifold of minimum loss. This is in contrast to many previous results about implicit bias either relying on infinitesimal updates or noise in gradient. Formally, for any smooth function $L$ with certain regularity condition, this effect is demonstrated for (1) Normalized GD, i.e., GD with a varying LR $η_t =\fracη{\| \nabla L(x(t)) \|}$ and loss $L$; (2) GD with constant LR and loss $\sqrt{L- \min_x L(x)}$. Both provably enter the Edge of Stability, with the associated flow on the manifold minimizing $λ_{1}(\nabla^2 L)$. The above theoretical results have been corroborated by an experimental study.
