98%
921
2 minutes
20
The recently proposed Kolmogorov-Arnold network (KAN) is a promising alternative to multi-layer perceptrons (MLPs) for data-driven modeling. While original KAN layers were only capable of representing the addition operator, the recently-proposed MultKAN layer combines addition and multiplication subnodes in an effort to improve representation performance. Here, we find that MultKAN layers suffer from a few key drawbacks including limited applicability in output layers, bulky parameterizations with extraneous activations, and the inclusion of complex hyperparameters. To address these issues, we propose LeanKANs, a direct and modular replacement for MultKAN and traditional AddKAN layers. LeanKANs address these three drawbacks of MultKAN through general applicability as output layers, significantly reduced parameter counts for a given network structure, and a smaller set of hyperparameters. As a one-to-one layer replacement for standard AddKAN and MultKAN layers, LeanKAN is able to provide these benefits to traditional KAN learning problems as well as augmented KAN structures in which it serves as the backbone, such as KAN Ordinary Differential Equations (KAN-ODEs) or Deep Operator KANs (DeepOKAN). We demonstrate LeanKAN's simplicity and efficiency in a series of demonstrations carried out across a standard KAN toy problem as well as ordinary and partial differential equations learned via KAN-ODEs, where we find that its sparser parameterization and compact structure serve to increase its expressivity and learning capability, leading it to outperform similar and even much larger MultKANs in various tasks.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1016/j.neunet.2025.107883 | DOI Listing |
IEEE Trans Neural Syst Rehabil Eng
September 2025
Sleep monitoring is essential for assessing sleep quality and understanding its broader implications for overall health. Although electroencephalography (EEG) remains the gold standard for sleep analysis, multichannel techniques are often cumbersome and impractical for real-world application. As a more feasible alternative, single-channel EEG offers greater practicality but still faces several persistent challenges, including reduced spatial resolution, feature instability, and limited clinical interpretability.
View Article and Find Full Text PDFSci Rep
September 2025
Data Analytics, Generation Australia, Sydney, 2000, Australia.
Carotid Intima-Media Thickness (CIMT) is defined as a non-invasive and well-validated sign of asymptomatic atherosclerosis and an early predictor of cardiovascular disease (CVD). We assembled a carefully curated dataset of 100 adult patients, encompassing 13 clinical, biochemical and demographic variables routinely collected in outpatient practice. After a five-stage pre-processing pipeline median/mode imputation, categorical encoding, Min-Max scaling, inter-quartile-range outlier removal and SMOTE-NC balancing we trained a Kolmogorov-Arnold Network (KAN) to assign each patient to one of four CIMT-defined risk tiers mentioned as "No", "Low", "Medium", "High".
View Article and Find Full Text PDFEnviron Monit Assess
August 2025
Reference Laboratory, Health Deputy, Iran University of Medical Sciences, Tehran, Iran.
In this study, the water quality index (WQI) was calculated using multivariate statistics, incorporating physical, chemical, and microbiological analysis of water samples taken from water supply networks in the western district of Tehran from 2021 to 2024. The principal drinking water parameters such as pH, total hardness, turbidity, lead (Pb), chloride (Cl), fluoride (F), total dissolved solids (TDS), sulfate (SO), nitrate (NO), nitrite (NO), calcium (Ca), magnesium (Mg), arsenic (As), mercury (Hg), cadmium (Cd), fecal coliform and total residual chlorine (Ch) were selected according to Iranian national water standards. The WQI index was predicted using various machine learning algorithms, including multiple linear regression (MLR), support vector machine (SVM) regression, extreme gradient boosting (XGBoost), Random Forest (RF) regression, multilayer perceptron (MLP), and Kolmogorov-Arnold networks (KAN).
View Article and Find Full Text PDFPlants (Basel)
August 2025
College of Computer & Mathematics, Central South University of Forestry and Technology, Changsha 410004, China.
High-quality disease segmentation plays a crucial role in the precise identification of rice diseases. Although the existing deep learning methods can identify the disease on rice leaves to a certain extent, these methods often face challenges in dealing with multi-scale disease spots and irregularly growing disease spots. In order to solve the challenges of rice leaf disease segmentation, we propose KBNet, a novel multi-modal framework integrating language and visual features for rice disease segmentation, leveraging the complementary strengths of CNN and Transformer architectures.
View Article and Find Full Text PDFSensors (Basel)
August 2025
Key Laboratory of Flight Techniques and Flight Safety, CAAC, Guanghan 618307, China.
Inspired by the interpretability of Kolmogorov-Arnold Networks (KANs), a novel Pixel-level Feature Selection (PFS) model based on KANs (PFSKANs) is proposed as a fundamentally distinct alternative from trainable Convolutional Neural Networks (CNNs) and transformers in the computer vision tasks. We modify the simplification techniques of KANs to detect key pixels with high contribution scores directly at the input image. Specifically, a trainable selection procedure is intuitively visualized and performed only once, since the obtained interpretable pixels can subsequently be identified and dimensionally standardized using the proposed mathematical approach.
View Article and Find Full Text PDF