Graph Signal Processing: Overview, Challenges, and Applications

By Antonio Ortega, Pascal Frossard, Jelena Kovacevic, José M. F. Moura, Pierre Vandergheynst. Proceedings of the IEEE, 2017

1,859 citations recorded by the source corpus.

Research in graph signal processing (GSP) aims to develop tools for processing data defined on irregular graph domains. In this paper, we first provide an overview of core ideas in GSP and their connection to conventional digital signal processing, along with a brief historical perspective to highlight how concepts recently developed in GSP build on top of prior research in other areas. We then summarize recent advances in developing basic GSP tools, including methods for sampling, filtering, or graph learning. Next, we review progress in several application areas using GSP, including processing and analysis of sensor network data, biological data, and applications to image processing and machine learning.

Subfields: Artificial Intelligence, Molecular Biology, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/36d442f59c61ea2912d227c24dee76778c546b0a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Molecular Biology

In Statistical and Nonlinear Physics

Citations in the corpus

Revisiting Graph Neural Networks: All We Have is Low-Pass Filters (2019, background): “SGC or GCN) fail to perform? In this study, we provide an answer to the aforementioned questions from the graph signal processing perspective [17].”
Graph Neural Networks for Decentralized Multi-Robot Path Planning (2019, background): “This vector xft is a graph signal [27], since it assigns a scalar value to each node, xft : V → R so that [x f t ]i = x if t ∈ R.”
Graph Representation Learning (2020, methodology): “There are also close theoretical connections between spectral clustering and random walks on graphs, as well as the field of graph signal processing Ortega et al. [2018]. We will discuss many of these connections in future chapters.”
Optimal Power Flow Using Graph Neural Networks (2019, background): “Matrices that satisfy this constraint are called graph shift operators (GSOs) and other examples of widespread use in the GSP literature include the Laplacian matrix, the random walk Laplacian and several normalized versions of these [26].”
Learning Graphs From Data: A Signal Representation Perspective (2018, methodology): “GSP enables the processing and analysis of signals that lie on structured but irregular domains by generalizing classical signal processing concepts, tools, and methods, such as time–frequency analysis and filtering, on graphs [3], [12], [13].”
Connecting the Dots: Identifying Network Structure via Graph Signal Processing (2018, background): “Coping with the challenges found at the intersection of Network Science and Big Data necessitates fundamental breakthroughs in modeling, identification, and controllability of distributed network processes – often conceptualized as signals defined on graphs [38].”
STREETS: A Novel Camera Network Dataset for Traffic Flow (2019, neutral): “[38] Lastly, computer vision has been explored as a mode of traffic analysis.”
LanczosNet: Multi-Scale Deep Graph Convolutional Networks (2019, methodology): “Built on top of the graph signal processing (GSP) approaches [4], these methods extend convolution operators to graphs by leveraging spectral graph theory, graph wavelet theory, etc. Graph convolutions can be stacked and combined with nonlinear activation functions to build deep models, just as in…”
Random Walks on Simplicial Complexes and the normalized Hodge Laplacian (2018, methodology): “This type of filtering is analogous to the filtering of signals defined on the nodes of a graph as considered in graph signal processing [96].”
Graph Learning: A Survey (2021, methodology): “Besides, GSP has also been proposed to deal with graph analysis [26].”