aiqbench analysis
SechKAN introduces a novel neural architecture using hyperbolic secant functions within the Kolmogorov-Arnold Network framework, demonstrating improved performance in function fitting and PDE solving while maintaining parameter efficiency comparable to standard MLPs. Although it offers stable gradients for scientific computing tasks, its inference speed remains slightly slower than traditional multilayer perceptrons despite being faster than other KAN variants. This research highlights an ongoing trade-off between model expressiveness/gradient stability and computational latency in next-generation neural network designs.
SechKAN: Kolmogorov-Arnold Networks with Hyperbolic Secant Functions
arXiv:2607.18290v1 Announce Type: cross Abstract: In recent years, Kolmogorov-Arnold Networks (KANs) have attracted increasing attention due to their effectiveness in machine learning and scientific computing tasks, offering a new paradigm for neural network design. In this paper, we present SechKAN, a KAN architecture based on hyperbolic secant (sech) functions. The hyperbolic secant basis is used for its smooth bell-shaped form, localized responses, and stable gradients. We employ 1D linear transformations to reduce the number of parameters, allowing SechKAN to remain comparable to multilayer perceptrons (MLPs) in model size. Experimental results indicate the effectiveness of SechKAN in function fitting, PDE problems, and image classification tasks on benchmark datasets, including MNIST, Fashion-MNIST, CIFAR-10, and CIFAR-100. SechKAN achieves superior performance compared to MLPs and other KAN variants while maintaining a similar number of parameters. However, its running time, while better than that of other KAN variants, is slightly longer than that of MLPs.
Abstract:In recent years, Kolmogorov-Arnold Networks (KANs) have attracted increasing attention due to their effectiveness in machine learning and scientific computing tasks, offering a new paradigm for neural network design. In this paper, we present SechKAN, a KAN architecture based on hyperbolic secant (sech) functions. The hyperbolic secant basis is used for its smooth bell-shaped form, localized responses, and stable gradients. We employ 1D linear transformations to reduce the number of parameters, allowing SechKAN to remain comparable to multilayer perceptrons (MLPs) in model size. Experimental results indicate the effectiveness of SechKAN in function fitting, PDE problems, and image classification tasks on benchmark datasets, including MNIST, Fashion-MNIST, CIFAR-10, and CIFAR-100. SechKAN achieves superior performance compared to MLPs and other KAN variants while maintaining a similar number of parameters. However, its running time, while better than that of other KAN variants, is slightly longer than that of MLPs.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2607.18290 [cs.LG]
(or arXiv:2607.18290v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2607.18290 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.
Full text under the publisher's open licence · 240 words
Related coverage
Evaluating the Effectiveness of SechKAN on 1D Data
arXiv · Sep 24, 2026
Hierarchical RBF-KAN and RBF-SKAN Architectures for Multidimensional Function Approximation and Random Field Learning
arXiv · Jun 3, 2026
Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning
arXiv · Oct 12, 2026
GS-KAN: Parameter-Efficient Kolmogorov-Arnold Networks via Sprecher-Type Shared Basis Functions
arXiv · Jun 8, 2026