AI RESEARCH
Geometric and Spectral Alignment for Deep Neural Network II
arXiv CS.LG
•
ArXi:2605.02111v1 Announce Type: new This paper develops the angular and static-channel component of Geometric and Spectral Alignment for residual Jacobian chains. Starting from Cartan-coordinate rigidity and fitted effective-rank windows, we study how dominant singular subspaces are transported across adjacent layers and how the resulting finite matrices can be displayed in physical channel coordinates. The main results are deterministic, margin-verified results.