Kolmogorov-Arnold Networks
PulseAugur coverage of Kolmogorov-Arnold Networks — every cluster mentioning Kolmogorov-Arnold Networks across labs, papers, and developer communities, ranked by signal.
- instance of alphaXiv 90%
- instance of ScienceCast 90%
- instance of Manchester Literary and Philosophical Society 90%
- instance of Gotit.pub 90%
- developed ScienceCast 90%
- instance of Influence Flower 90%
- instance of CatalyzeX 90%
- developed CatalyzeX Code Finder for Papers 90%
- instance of Chebyshev 90%
- instance of Kolmogorov-Arnold representation theorem 90%
- competes with multilayer perceptron 70%
- used by alphaXiv 70%
- 2026-08-12 research_milestone A new method, FS-JEPA, was proposed to improve KANs for medical image segmentation, achieving state-of-the-art results. source
11 day(s) with sentiment data
-
TorchCurves library enables learnable parametric curves in PyTorch
A new PyTorch library called TorchCurves has been released, enabling vector-valued parametric curves to be treated as learnable primitives within neural networks. The library supports curves represented by B-spline or L…
-
New Bounds Reveal Kolmogorov-Arnold Networks' Geometric Complexity Advantage
Researchers have established tight asymptotic bounds on the maximum number of linear regions for continuous piecewise-linear Kolmogorov-Arnold Networks (KANs). These bounds, which include both upper and lower limits, de…
-
MetaCircle secures funding for AI "intuition" models based on KAN
MetaCircle, a startup founded by a Tsinghua University assistant professor, has secured several hundred million yuan in funding across two rounds within four months. The company is developing "Meta Models," which aim to…
-
New GeoDiv KAN architecture advances fluid dynamics simulations
Researchers have developed a new machine learning architecture called GeoDiv KAN to solve complex fluid dynamics problems. This model, which incorporates geometry-dependent feature gating and a streamfunction formulatio…
-
New S2KAN method enhances interpretability of Kolmogorov-Arnold Networks
Researchers have introduced Softly Symbolified Kolmogorov-Arnold Networks (S2KAN), a novel approach to enhance the interpretability of machine learning models. S2KAN integrates symbolic primitives directly into the trai…
-
New neural activation function improves edge AI efficiency
Researchers have introduced a new neural network architecture that aims to improve computational efficiency and hardware scalability for edge devices. This framework utilizes a novel "Parametric Tanh Connection Function…
-
New I2-KAN framework improves multimodal sentiment analysis interpretation
Researchers have developed I$^2$-KAN, a novel framework for multimodal sentiment analysis that disentangles intra-modal and inter-modal information pathways. This approach uses a Task-Decoupled Linear-Gated Denoising Bo…
-
New TOUCKANs architecture enables tractable inference for scientific AI
Researchers have introduced Kolmogorov-Arnold Networks with Tractable Operators for Uncertainty and Constraints (TOUCKANs), a novel neural network architecture designed for computational science. Unlike conventional net…
-
New KAN-based method enhances NLP embedding layer compression
Researchers have developed a novel method for compressing embedding layers in natural language processing models by incorporating non-linear projections inspired by Kolmogorov-Arnold Networks (KANs). This approach aims …
-
New OOD detection framework for continual learning leverages KAN
Researchers have developed a new framework for out-of-distribution detection in continual learning (OOD-CL) that addresses the challenges of learning sequential tasks without forgetting and identifying unknown samples w…
-
New SUPN architecture offers efficient function approximation
Researchers have introduced Shallow Universal Polynomial Networks (SUPNs), a novel architecture designed for efficient function approximation. Unlike deep neural networks (DNNs) and Kolmogorov-Arnold Networks (KANs) tha…
-
Kolmogorov-Arnold Networks: Benefits questioned, educational tool released
A new paper questions the universal benefits of Kolmogorov-Arnold Networks (KANs), suggesting their favorable scaling properties may be tied to specific target structures rather than an inherent architectural advantage.…
-
New AI Model HankelKANet Improves Wireless Path Loss Estimation
Researchers have developed HankelKANet, a novel neural network designed for efficient and accurate path loss estimation in wireless networks. This lightweight model integrates physics-inspired principles, specifically u…
-
KANLab offers interactive browser tool for learning Kolmogorov-Arnold Networks
KANLab is a new open-source, browser-based tool designed to help users learn about Kolmogorov-Arnold Networks (KANs). It visualizes B-spline activation curves, gradient descent, and pruning in a simplified 2D environmen…
-
New INR framework fuses MLP and KAN for improved signal representation
Researchers have developed a new Implicit Neural Representation (INR) framework that fuses a Multi-Layer Perceptron (MLP) with a Kolmogorov-Arnold Network (KAN) to better model multidimensional signals. This hybrid appr…
-
New mesh-based neural energy method improves material simulation accuracy
Researchers have introduced a mesh-based neural energy method (M-NEM) to improve the simulation of heterogeneous materials. This new approach addresses limitations of existing physics-informed neural networks, such as t…
-
Kolmogorov-Arnold Networks show significant sample-efficiency gains in RL
A new study published on arXiv explores the sample-efficiency of Kolmogorov-Arnold Networks (KANs) in deep reinforcement learning. The research demonstrates that KANs can achieve comparable performance to traditional mu…
-
New KANSteer method models nonlinear concept traversal in text-to-image models
Researchers have proposed KANSteer, a new method for interpreting text-to-image models that moves beyond the Linear Representation Hypothesis. This hypothesis assumes concepts are encoded as linear directions in activat…
-
Kolmogorov-Arnold Networks show promise in scaling, PDEs, and interpretability · 3 sources tracked
Researchers are exploring the capabilities and theoretical underpinnings of Kolmogorov-Arnold Networks (KANs), an alternative to traditional Multi-Layer Perceptrons (MLPs). One study investigates KANs' neural scaling la…
-
New physics-informed method adaptively constructs functional representations for complex equations
Researchers have developed a novel Physics-Informed Method of Group Data Handling (PI-GMDH) that adaptively constructs functional representations for solving complex physical equations. This method progressively builds …