Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Graph Collaborative Filtering with FourierKAN Feature Transformation

Jinfeng Xu, Zheyu Chen, Jinze Li, Shuo Yang, Wei Wang, Xiping Hu, Edith Ngai

TL;DR

This work reevaluates the role of feature transformation in graph-based collaborative filtering and introduces FourierKAN-GCF, which replaces traditional MLP-based transformations with Fourier Kolmogorov-Arnold Networks to enhance representation while easing training. By removing the need for heavy transform matrices and employing a Fourier-coefficient-based activation, the model achieves superior performance over common GCF backbones and remains compatible with advanced self-supervised backbones. Ablations and sensitivity analyses demonstrate the effectiveness and tunability of FourierKAN, particularly through the grid size g and dropout strategies. The proposed approach offers a practical, scalable path to stronger graph-based recommendations across diverse datasets.

Abstract

Graph Collaborative Filtering (GCF) has emerged as a dominant paradigm in modern recommendation systems, excelling at modeling complex user-item interactions and capturing high-order collaborative signals through graph-structured learning. Most existing GCF models predominantly rely on simplified graph architectures like LightGCN, which strategically remove feature transformation and activation functions from vanilla graph convolution networks. Through systematic analysis, we reveal that feature transformation in message propagation can enhance model representation, though at the cost of increased training difficulty. To this end, we propose FourierKAN-GCF, a novel GCN framework that adopts Fourier Kolmogorov-Arnold Networks as efficient transformation modules within graph propagation layers. This design enhances model representation while decreasing training difficulty. Our FourierKAN-GCF can achieve higher recommendation performance than most widely used GCF backbone models. In addition, it can be integrated into existing advanced self-supervised models as a backbone, replacing their original backbone to achieve enhanced performance. Extensive experiments on three public datasets demonstrate the superiority of FourierKAN-GCF.

Enhancing Graph Collaborative Filtering with FourierKAN Feature Transformation

TL;DR

This work reevaluates the role of feature transformation in graph-based collaborative filtering and introduces FourierKAN-GCF, which replaces traditional MLP-based transformations with Fourier Kolmogorov-Arnold Networks to enhance representation while easing training. By removing the need for heavy transform matrices and employing a Fourier-coefficient-based activation, the model achieves superior performance over common GCF backbones and remains compatible with advanced self-supervised backbones. Ablations and sensitivity analyses demonstrate the effectiveness and tunability of FourierKAN, particularly through the grid size g and dropout strategies. The proposed approach offers a practical, scalable path to stronger graph-based recommendations across diverse datasets.

Abstract

Graph Collaborative Filtering (GCF) has emerged as a dominant paradigm in modern recommendation systems, excelling at modeling complex user-item interactions and capturing high-order collaborative signals through graph-structured learning. Most existing GCF models predominantly rely on simplified graph architectures like LightGCN, which strategically remove feature transformation and activation functions from vanilla graph convolution networks. Through systematic analysis, we reveal that feature transformation in message propagation can enhance model representation, though at the cost of increased training difficulty. To this end, we propose FourierKAN-GCF, a novel GCN framework that adopts Fourier Kolmogorov-Arnold Networks as efficient transformation modules within graph propagation layers. This design enhances model representation while decreasing training difficulty. Our FourierKAN-GCF can achieve higher recommendation performance than most widely used GCF backbone models. In addition, it can be integrated into existing advanced self-supervised models as a backbone, replacing their original backbone to achieve enhanced performance. Extensive experiments on three public datasets demonstrate the superiority of FourierKAN-GCF.
Paper Structure (20 sections, 6 equations, 3 figures, 4 tables)

This paper contains 20 sections, 6 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Preliminary
  3. NGCF Brief
  4. LightGCN Brief
  5. Re-Analysis for NGCF
  6. Methodology
  7. Kolmogorov-Arnold Network (KAN)
  8. Fourier KAN
  9. FourierKAN-GCF
  10. Dropout Strategies
  11. Model Training
  12. Experiment
  13. Datasets and Evaluation Metrics
  14. Baselines and Experimental Settings
  15. Performance Comparison
  16. ...and 5 more sections

Figures (3)

  • Figure 1: Overview of FourierKAN-GCF.
  • Figure 2: Effective of layer number $L$ and grid size $g$.
  • Figure 3: Study of dropout ratio pair $p_m$ and $p_n$ (Recall@20).