arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.06208cs.LGcs.AI
收藏

多层感知机中平滑激活函数下的低秩训练动态的涌现

Emergent Low-Rank Training Dynamics in MLPs with Smooth Activations

University of Michigan(密歇根大学);EECS Department, University of Michigan, Ann Arbor, MI, USA(密歇根大学电子工程与计算机科学系)

Alec S. Xu, Can Yaras, Matthew Asato, Qing Qu, Laura Balzano

AI总结:

本文研究了多层感知机中平滑激活函数下的低秩训练动态,揭示了权重动态在低维子空间中的集中现象,并展示了低秩参数化在分类任务中的有效性。

摘要:

最近的实证证据表明,大规模深度神经网络的训练动态发生在低维子空间中。尽管这激发了对低秩训练、压缩和适应的新研究,但在非线性网络中这些动态的理论依据仍然有限。为了填补这一空白,本文分析了在梯度下降(GD)下多层感知机(MLPs)的学习动态。我们证明了权重动态在整个训练过程中集中在不变的低维子空间中。理论上,我们精确地表征了具有平滑非线性激活函数的两层网络的这些不变子空间,提供了其出现的见解。实验上,我们验证了这种现象超出了我们的理论假设。利用这些见解,我们实证地展示了存在一种低秩MLP参数化方式,当在适当的子空间中初始化时,可以在多种分类任务上与完全参数化模型达到相同的分类性能。

英文摘要:

Recent empirical evidence has demonstrated that the training dynamics of large-scale deep neural networks occur within low-dimensional subspaces. While this has inspired new research into low-rank training, compression, and adaptation, theoretical justification for these dynamics in nonlinear networks remains limited. %compared to deep linear settings. To address this gap, this paper analyzes the learning dynamics of multi-layer perceptrons (MLPs) under gradient descent (GD). We demonstrate that the weight dynamics concentrate within invariant low-dimensional subspaces throughout training. Theoretically, we precisely characterize these invariant subspaces for two-layer networks with smooth nonlinear activations, providing insight into their emergence. Experimentally, we validate that this phenomenon extends beyond our theoretical assumptions. Leveraging these insights, we empirically show there exists a low-rank MLP parameterization that, when initialized within the appropriate subspaces, matches the classification performance of fully-parameterized counterparts on a variety of classification tasks.

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。

↑