PAKAN: Pixel Adaptive Kolmogorov-Arnold Network Modules for Pansharpening

2026-03-16

AI Summary

This work addresses the limitation of existing deep networks in pan-sharpening remote sensing imagery, which rely on static activation functions and struggle to model complex spatial-spectral nonlinear mappings. To overcome this, the study introduces, for the first time, a pixel-wise adaptive mechanism into Kolmogorov–Arnold Networks (KANs), proposing two novel modules—PAKAN 2to1 and PAKAN 1to1—that dynamically generate spline summation weights along spatial and spectral dimensions. This enables local adaptation of activation behavior, surpassing the constraints of conventional static or globally learnable activations. Extensive experiments on multiple benchmarks demonstrate that the proposed approach significantly outperforms state-of-the-art methods, validating the effectiveness and superiority of pixel-adaptive activation in pan-sharpening.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Active Learning
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Pansharpening aims to fuse high-resolution spatial details from panchromatic images with the rich spectral information of multispectral images. Existing deep neural networks for this task typically rely on static activation functions, which limit their ability to dynamically model the complex, non-linear mappings required for optimal spatial-spectral fusion. While the recently introduced Kolmogorov-Arnold Network (KAN) utilizes learnable activation functions, traditional KANs lack dynamic adaptability during inference. To address this limitation, we propose a Pixel Adaptive Kolmogorov-Arnold Network framework. Starting from KAN, we design two adaptive variants: a 2D Adaptive KAN that generates spline summation weights across spatial dimensions and a 1D Adaptive KAN that generates them across spectral channels. These two components are then assembled into PAKAN 2to1 for feature fusion and PAKAN 1to1 for feature refinement. Extensive experiments demonstrate that our proposed modules significantly enhance network performance, proving the effectiveness and superiority of pixel-adaptive activation in pansharpening tasks.

Problem

• pansharpening
• Kolmogorov-Arnold Network
• adaptive activation
• spatial-spectral fusion
• deep neural networks

Innovation

• Pixel Adaptive
• Kolmogorov-Arnold Network
• Pansharpening
• Learnable Activation Functions
• Spatial-Spectral Fusion