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 2406.07712cs.LG
收藏

基于损失梯度高斯宽度的泛化与优化保证

Loss Gradient Gaussian Width based Generalization and Optimization Guarantees

Siebel School of Computing and Data Science(西贝尔计算与数据科学学院);University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);University of Minnesota, Twin Cities(明尼苏达大学双城分校)

Arindam Banerjee, Qiaobo Li, Yingxue Zhou

AI总结:

本文提出以损失梯度高斯宽度(LGGW)度量梯度复杂度,在包含 PL 条件的梯度支配条件下建立泛化与优化保证,并证明小 LGGW 下样本重用不会使经验梯度偏离总体梯度,同时用特征提取器高斯宽度界定深度网络的单样本 LGGW,为深度模型紧致界提供新途径。

摘要:

关于总体损失的泛化与优化保证通常依赖于基于一致收敛的分析,通常以预测器的 Rademacher 复杂度为基础。现代模型强大的表示能力引发了人们对这一方法的担忧。在本文中,我们以梯度复杂度为度量,即损失梯度高斯宽度(Loss Gradient Gaussian Width,LGGW),给出泛化与优化保证。首先,我们在灵活的梯度支配条件下直接以 LGGW 给出泛化保证,该条件包含流行的 PL(Polyak-Łojasiewicz)条件作为特例。其次,我们证明,只要 LGGW 较小,迭代梯度下降中的样本重用不会使经验梯度偏离总体梯度。第三,针对深度网络,我们将单样本 LGGW 用特征提取器(即倒数第二层的输出)的高斯宽度进行上界估计。据我们所知,我们以 LGGW 形式给出的泛化与优化保证是同类结果中的首个成果,并且有望为深度模型带来定量上紧致的界。

英文摘要:

Generalization and optimization guarantees on the population loss often rely on uniform convergence based analysis, typically based on the Rademacher complexity of the predictors. The rich representation power of modern models has led to concerns about this approach. In this paper, we present generalization and optimization guarantees in terms of the complexity of the gradients, as measured by the Loss Gradient Gaussian Width (LGGW). First, we introduce generalization guarantees directly in terms of the LGGW under a flexible gradient domination condition, which includes the popular PL (Polyak-Łojasiewicz) condition as a special case. Second, we show that sample reuse in iterative gradient descent does not make the empirical gradients deviate from the population gradients as long as the LGGW is small. Third, focusing on deep networks, we bound their single-sample LGGW in terms of the Gaussian width of the featurizer, i.e., the output of the last-but-one layer. To our knowledge, our generalization and optimization guarantees in terms of LGGW are the first results of its kind, and hold considerable promise towards quantitatively tight bounds for deep models.

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。

↑