Graph Signal Processing (GSP)

2024-07-12 167 words One minute

Contents

Graph Signal Processing (GSP) is a mathematical framework that extends discrete signal processing (DSP) tools such as filtering and signal decomposition to graph data structures.

Graph Signal Processing is suitable for scenarios where data don’t have a regular pattern or cannot be represented as time series signals. In such scenarios, modeling through graph structures can exploit signal-processing tasks on complex structures through GSP.

GSP can be applied to network structures such as sensor, social, and biological networks. As well as image processing.

A GSP implementation as a Toolbox for Matlab and Python is available. In it, you can build graph structures and apply signal processing tasks like filtering.

/posts/gsp/original-signal.png — Original Signal

/posts/gsp/noisy-signal.png — Noisy Signal

/posts/gsp/denoisy-signal.png — Denoised Signal

Ortega and his friends (2018) provided an overview of GSP, highlighting its application areas and use cases. The publication and frameworks provide more information about GSP.

Note

Ortega, A., Frossard, P., Kovačević, J., Moura, J. M., & Vandergheynst, P. (2018). Graph signal processing: Overview, challenges, and applications. Proceedings of the IEEE, 106(5), 808–828.