aiqbench analysis
This research introduces HermNet, a spectral graph neural network leveraging Hermite polynomials that avoids computationally expensive eigendecomposition through sparse recurrence relations. The study provides a rigorous analysis of optimization dynamics under limited training budgets and demonstrates how specific regularization techniques can enhance performance in short-horizon scenarios compared to other polynomial bases. These findings offer practical guidance for researchers selecting basis functions and architectural constraints when designing efficient graph learning models.
Spectral Graph Neural Networks with Hermite Polynomials: A Comprehensive Study
arXiv:2609.28979v1 Announce Type: new Abstract: We study spectral graph neural networks built from Hermite polynomials and propose HermNet, a simple model that combines a nodewise predictor with normalized Hermite propagation. Its sparse recurrence requires neither eigendecomposition nor a learned basis. We distinguish the basic model from optional coordinate calibration, response normalization and Gaussian derivative regularization. Hermite and other complete polynomial bases span the same degree-bounded filter space, but their coordinates can produce different optimization behavior under limited training budgets. We analyze this behavior through spectral signal energy, label sampling, changes in learned features and the bias--variance trade-off of regularization. Controlled synthetic experiments identify a regime in which plain HermNet outperforms matched polynomial-basis alternatives, including with a jointly trained nonlinear predictor. Curvature regularization further improves HermNet when the same functional penalty is available to every comparator. Fixed-predictor controls support the advantage under short training budgets, but longer training removes the plain-model lead. Matched real-data comparisons show accuracy deficits, and architectural and numerical studies identify further limits. Together, the analysis and experiments clarify when Hermite propagation is useful and how calibration and regularization affect its performance.
Abstract:We study spectral graph neural networks built from Hermite polynomials and propose HermNet, a simple model that combines a nodewise predictor with normalized Hermite propagation. Its sparse recurrence requires neither eigendecomposition nor a learned basis. We distinguish the basic model from optional coordinate calibration, response normalization and Gaussian derivative regularization. Hermite and other complete polynomial bases span the same degree-bounded filter space, but their coordinates can produce different optimization behavior under limited training budgets. We analyze this behavior through spectral signal energy, label sampling, changes in learned features and the bias--variance trade-off of regularization. Controlled synthetic experiments identify a regime in which plain HermNet outperforms matched polynomial-basis alternatives, including with a jointly trained nonlinear predictor. Curvature regularization further improves HermNet when the same functional penalty is available to every comparator. Fixed-predictor controls support the advantage under short training budgets, but longer training removes the plain-model lead. Matched real-data comparisons show accuracy deficits, and architectural and numerical studies identify further limits. Together, the analysis and experiments clarify when Hermite propagation is useful and how calibration and regularization affect its performance.
Machine Learning (cs.LG)
arXiv:2609.28979 [cs.LG]
(or arXiv:2609.28979v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.28979 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 273 words
Related coverage
ChronoSheaf: Joint Spatio-Temporal Spectral Filtering on Cellular Sheaves
arXiv · Oct 16, 2026
LiftGCN: Efficient Energy-Preserving Graph Learning via Joukowski Spectral Lifting for Finite Element Stress Prediction
arXiv · Sep 16, 2026
Nonlinear Laplacians Improve Signed-Directed Graph Learning
arXiv · Aug 4, 2026
A Unified Uncertainty Representation for Graph Neural Networks via Doubly-Spectral Stochastic Expansion
arXiv · Sep 30, 2026