{"id":"arxiv/2608.16180","url":"https://arxiv.org/abs/2608.16180","tags":["cs.LG","math.DG"],"score":0,"title":"Demystifying Oversmoothing in Sheaf Neural Networks: An Index-Theoretic Criterion","domain":"cs.LG","source":"arxiv","authors":["Junwen Dong","Yuhan Peng","Hao Li","Huitao Feng","Kelin Xia"],"reviews":0,"subject":"cs.LG","summary":"To combat oversmoothing in Graph Convolutional Networks, Sheaf Neural Networks (SNNs) were proposed as a generalization by equipping the graph with a sheaf structure and replacing the graph Laplacian with a sheaf Laplacian $\\mathcal{L}$. Existing analyses connect sheaf diffusion to oversmoothing via the harmonic space ($\\ker\\mathcal{L}$), taking its absolute dimension as an indicator of anti-oversmoothing capacity. However, absolute dimension alone is not a reliable measure: certain sheaf configurations inflate $\\dim \\ker \\mathcal{L}$ while their harmonic sections remain entirely constant,…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-08-17","reactions":{}}