Rotation-Parameterized Graph Fractional Fourier Transform: Definition, Properties, and Optimal Filtering
Researchers have introduced a new method called the rotation-parameterized graph fractional Fourier transform (RP-GFRFT) to enhance the analysis of graph-structured data. This novel approach unifies fractional order and rotation-parameterized spectral analysis, ensuring theoretical consistency by guaranteeing reduction to the standard graph Fourier transform at zero angle. Experiments show that RP-GFRFT outperforms existing methods in denoising, reconstruction, and feature preservation for real-world signals, images, and point clouds. AI
IMPACT Introduces a novel transform for more effective analysis of graph-structured data, potentially improving performance in AI tasks involving such data.