On the expressiveness and spectral bias of KANs

• URL: http://arxiv.org/abs/2410.01803v2
• Date: Thu, 06 Feb 2025 19:49:32 GMT
• Title: On the expressiveness and spectral bias of KANs
• Authors: Yixuan Wang, Jonathan W. Siegel, Ziming Liu, Thomas Y. Hou,
• Abstract summary: KANs were recently proposed as a potential alternative to the prevalent architectural backbone of many deep learning models, the multi-layer perceptron (MLP) KANs have seen success in various tasks of AI for science, with their empirical efficiency and accuracy demostrated in function regression, PDE solving, and many more scientific problems.
• Score: 17.42614039265962
• License:
• Abstract: Kolmogorov-Arnold Networks (KAN) \cite{liu2024kan} were very recently proposed as a potential alternative to the prevalent architectural backbone of many deep learning models, the multi-layer perceptron (MLP). KANs have seen success in various tasks of AI for science, with their empirical efficiency and accuracy demostrated in function regression, PDE solving, and many more scientific problems. In this article, we revisit the comparison of KANs and MLPs, with emphasis on a theoretical perspective. On the one hand, we compare the representation and approximation capabilities of KANs and MLPs. We establish that MLPs can be represented using KANs of a comparable size. This shows that the approximation and representation capabilities of KANs are at least as good as MLPs. Conversely, we show that KANs can be represented using MLPs, but that in this representation the number of parameters increases by a factor of the KAN grid size. This suggests that KANs with a large grid size may be more efficient than MLPs at approximating certain functions. On the other hand, from the perspective of learning and optimization, we study the spectral bias of KANs compared with MLPs. We demonstrate that KANs are less biased toward low frequencies than MLPs. We highlight that the multi-level learning feature specific to KANs, i.e. grid extension of splines, improves the learning process for high-frequency components. Detailed comparisons with different choices of depth, width, and grid sizes of KANs are made, shedding some light on how to choose the hyperparameters in practice.

Related papers

• Low Tensor-Rank Adaptation of Kolmogorov--Arnold Networks [70.06682043272377]
  Kolmogorov--Arnold networks (KANs) have demonstrated their potential as an alternative to multi-layer perceptions (MLPs) in various domains. We develop low tensor-rank adaptation (LoTRA) for fine-tuning KANs. We explore the application of LoTRA for efficiently solving various partial differential equations (PDEs) by fine-tuning KANs.
  arXiv  Detail & Related papers  (2025-02-10T04:57:07Z)
• PRKAN: Parameter-Reduced Kolmogorov-Arnold Networks [47.947045173329315]
  Kolmogorov-Arnold Networks (KANs) represent an innovation in neural network architectures. KANs offer a compelling alternative to Multi-Layer Perceptrons (MLPs) in models such as CNNs, RecurrentReduced Networks (RNNs) and Transformers. This paper introduces PRKANs, which employ several methods to reduce the parameter count in layers, making them comparable to Neural M layers.
  arXiv  Detail & Related papers  (2025-01-13T03:07:39Z)
• PowerMLP: An Efficient Version of KAN [10.411788782126091]
  The Kolmogorov-Arnold Network (KAN) is a new network architecture known for its high accuracy in several tasks such as function fitting and PDE solving. The superior computation capability of KAN arises from the Kolmogorov-Arnold representation and learnable spline functions. PowerMLP achieves higher accuracy and a training speed about 40 times faster than KAN in various tasks.
  arXiv  Detail & Related papers  (2024-12-18T07:42:34Z)
• Incorporating Arbitrary Matrix Group Equivariance into KANs [69.30866522377694]
  We propose Equivariant Kolmogorov-Arnold Networks (EKAN), a method for incorporating arbitrary matrix group equivariants into KANs. EKAN achieves higher accuracy with smaller datasets or fewer parameters on symmetry-related tasks, such as particle scattering and the three-body problem.
  arXiv  Detail & Related papers  (2024-10-01T06:34:58Z)
• A preliminary study on continual learning in computer vision using Kolmogorov-Arnold Networks [43.70716358136333]
  Kolmogorov- Networks (KAN) are based on a fundamentally different mathematical framework. KANs address several major issues insio, such as forgetting in continual learning scenarios. We extend the investigation by evaluating the performance of KANs in continual learning tasks within computer vision.
  arXiv  Detail & Related papers  (2024-09-20T14:49:21Z)
• KAN v.s. MLP for Offline Reinforcement Learning [4.3621896506713185]
  Kolmogorov-Arnold Networks (KAN) is an emerging neural network architecture in machine learning. In this paper, we explore the incorporation of KAN into the actor and critic networks for offline reinforcement learning.
  arXiv  Detail & Related papers  (2024-09-15T07:52:44Z)
• Kolmogorov-Arnold Network for Online Reinforcement Learning [0.22615818641180724]
  Kolmogorov-Arnold Networks (KANs) have shown potential as an alternative to Multi-Layer Perceptrons (MLPs) in neural networks. KANs provide universal function approximation with fewer parameters and reduced memory usage.
  arXiv  Detail & Related papers  (2024-08-09T03:32:37Z)
• KAN or MLP: A Fairer Comparison [63.794304207664176]
  This paper offers a fairer and more comprehensive comparison of KAN and models across various tasks. We control the number of parameters and FLOPs to compare the performance of KAN and representation. We find that KAN's issue is more severe than that of forgetting in a standard class-incremental continual learning setting.
  arXiv  Detail & Related papers  (2024-07-23T17:43:35Z)
• KAN: Kolmogorov-Arnold Networks [16.782018138008578]
  We propose Kolmogorov-Arnold Networks (KANs) as promising alternatives to Multi-Layer Perceptrons (MLPs) KANs have learnable activation functions on edges ("weights") We show that this seemingly simple change makes KANs outperform neurals in terms of accuracy and interpretability.
  arXiv  Detail & Related papers  (2024-04-30T17:58:29Z)
• ReLU Fields: The Little Non-linearity That Could [62.228229880658404]
  We investigate what is the smallest change to grid-based representations that allows for retaining the high fidelity result ofs. We show that such an approach becomes competitive with the state-of-the-art.
  arXiv  Detail & Related papers  (2022-05-22T13:42:31Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.