Demystifying Lazy Training of Neural Networks from a Macroscopic Viewpoint

arXiv:2404.04859v1 Announce Type: new
Abstract: In this paper, we advance the understanding of neural network training dynamics by examining the intricate interplay of various factors introduced by weight parameters in the initialization process. Motivated by the foundational work of Luo et al. (J. Mach. Learn. Res., Vol. 22, Iss. 1, No. 71, pp 3327-3373), we explore the gradient descent dynamics of neural networks through the lens of macroscopic limits, where we analyze its behavior as width $m$ tends to infinity. …

abstract advance arxiv cs.lg dynamics foundational iss lazy learn network networks network training neural network neural networks paper parameters process stat.ml training type understanding work