analiza aiqbench
This research introduces ShapKAN, a novel pruning method for Kolmogorov-Arnold Networks (KANs) that uses Shapley values to ensure node importance rankings remain consistent despite input coordinate shifts. By addressing the sensitivity of conventional magnitude-based methods in KAN architectures, this work enhances model interpretability and enables efficient compression without sacrificing symbolic representation recovery. The approach advances machine learning by bridging explainable AI techniques with emerging neural network structures designed for functional discovery.
Shift-Invariant Attribute Scoring for Kolmogorov-Arnold Networks via Shapley Value
arXiv:2510.01663v2 Announce Type: replace-cross Abstract: For many real-world applications, understanding feature-outcome relationships is as crucial as achieving high predictive accuracy. While traditional neural networks excel at prediction, their black-box nature obscures underlying functional relationships. Kolmogorov--Arnold Networks (KANs) address this by employing learnable spline-based activation functions on edges, enabling recovery of symbolic representations while maintaining competitive performance. However, KAN's architecture presents unique challenges for network pruning. Conventional magnitude-based methods become unreliable due to sensitivity to input coordinate shifts. We propose \textbf{ShapKAN}, a pruning framework using Shapley value attribution to assess node importance in a shift-invariant manner. Unlike magnitude-based approaches, ShapKAN quantifies each node's actual contribution, ensuring consistent importance rankings regardless of input parameterization. Extensive experiments on synthetic and real-world datasets demonstrate that ShapKAN preserves true node importance while enabling effective network compression. Our approach improves KAN's interpretability advantages, facilitating deployment in resource-constrained environments.
Abstract:For many real-world applications, understanding feature-outcome relationships is as crucial as achieving high predictive accuracy. While traditional neural networks excel at prediction, their black-box nature obscures underlying functional relationships. Kolmogorov--Arnold Networks (KANs) address this by employing learnable spline-based activation functions on edges, enabling recovery of symbolic representations while maintaining competitive performance. However, KAN's architecture presents unique challenges for network pruning. Conventional magnitude-based methods become unreliable due to sensitivity to input coordinate shifts. We propose \textbf{ShapKAN}, a pruning framework using Shapley value attribution to assess node importance in a shift-invariant manner. Unlike magnitude-based approaches, ShapKAN quantifies each node's actual contribution, ensuring consistent importance rankings regardless of input parameterization. Extensive experiments on synthetic and real-world datasets demonstrate that ShapKAN preserves true node importance while enabling effective network compression. Our approach improves KAN's interpretability advantages, facilitating deployment in resource-constrained environments.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2510.01663 [cs.LG]
(or arXiv:2510.01663v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2510.01663 Focus to learn more arXiv-issued DOI via DataCite
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.
Text integral sub licența deschisă a publicației · 232 cuvinte
Subiecte conexe
How do neural networks regress cubic polynomials? Apparently, they use a trick invented in Milan 500 years ago
LessWrong · 22 iul. 2026
NObSP: Functional Decomposition of Neural Networks via Oblique Subspace Projections
arXiv · 17 sept. 2026
Functional anatomy of Pythia-Herwig differences with Kolmogorov-Arnold networks
arXiv · 19 aug. 2026
Revalidation Beats Stateful Routing for Scientific Surrogates Under Distribution Shift
arXiv · 26 sept. 2026