In this work, we are interested in generalizing convolutional neural networks (CNNs) from low-dimensional regular grids, where image, video and speech are represented, to high-dimensional irregular domains, such as social networks, brain connectomes or words' embedding, represented by graphs. We present a formulation of CNNs in the context of spectral graph theory, which provides the necessary mathematical background and efficient numerical schemes to design fast localized convolutional filters on graphs. Importantly, the proposed technique offers the same linear computational complexity and constant learning complexity as classical CNNs, while being universal to any graph structure. Experiments on MNIST and 20NEWS demonstrate the ability of this novel deep learning system to learn local, stationary, and compositional features on graphs.

Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

/pdf/convolutional-neural-networks-on-graphs-with-fast-localized-1rjlnh7c4h.pdf

Convolutional neural networks on graphs with fast localized spectral filtering

Information Theory and Reliable Communication

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.

Graph Signal Processing: Overview, Challenges, and Applications

Fundamental theory: Holomorphic functions and domains of holomorphy Implicit functions and analytic sets The Poincare, Cousin, and Runge problems Pseudoconvex domains and pseudoconcave sets Holomorphic mappings Theory of analytic spaces: Ramified domains Analytic sets and holomorphic functions Analytic spaces Normal pseudoconvex spaces Bibliography Index.

Function theory in several complex variables

In many applications, the observations can be represented as a signal defined over the vertices of a graph. The analysis of such signals requires the extension of standard signal processing tools. In this paper, first, we provide a class of graph signals that are maximally concentrated on the graph domain and on its dual. Then, building on this framework, we derive an uncertainty principle for graph signals and illustrate the conditions for the recovery of band-limited signals from a subset of samples. We show an interesting link between uncertainty principle and sampling and propose alternative signal recovery algorithms, including a generalization to frame-based reconstruction methods. After showing that the performance of signal recovery algorithms is significantly affected by the location of samples, we suggest and compare a few alternative sampling strategies. Finally, we provide the conditions for perfect recovery of a useful signal corrupted by sparse noise, showing that this problem is also intrinsically related to vertex-frequency localization properties.

Signals on Graphs: Uncertainty Principle and Sampling

Developing tools to analyze signals defined over a graph is a research area that is attracting a significant amount of contributions because of its many applications. However, a graph representation does not capture the overall information about the data, as it implicitly takes into account only pairwise relations. The goal of this paper is to extend signal processing tools to signals defined over hypergraphs, which represent a formal framework to describe multi-way relations among the data. First, we suggest alternative ways to introduce a Fourier Transform (FT) for signals defined over hypergraphs and, in particular, for simplicial complexes. Then, building on the notion of Fourier Transform, we derive a sampling theorem aimed at identifying the minimum number of samples necessary to encode all information about band-limited hypergraph signals.

An introduction to hypergraph signal processing

Continuous-time signals are well known for not being perfectly localized in both time and frequency domains. Conversely, a signal defined over the vertices of a graph can be perfectly localized in both vertex and frequency domains. We derive the conditions ensuring the validity of this property and then, building on this theory, we provide the conditions for perfect reconstruction of a graph signal from its samples. Next, we provide a finite step algorithm for the reconstruction of a band-limited signal from its samples and then we show the effect of sampling a non perfectly band-limited signal and show how to select the bandwidth that minimizes the mean square reconstruction error.

/pdf/on-the-degrees-of-freedom-of-signals-on-graphs-1y5uffgi2m.pdf

On the degrees of freedom of signals on graphs

In many applications of current interest, the observations are represented as a signal defined over a graph. The analysis of such signals requires the extension of standard signal processing tools. Building on the recently introduced Graph Fourier Transform, the first contribution of this paper is to provide an uncertainty principle for signals on graph. As a by-product of this theory, we show how to build a dictionary of maximally concentrated signals on vertex/frequency domains. Then, we establish a direct relation between uncertainty principle and sampling, which forms the basis for a sampling theorem of signals defined on graph. Based on this theory, we show that, besides sampling rate, the samples' location plays a key role in the performance of signal recovery algorithms. Hence, we suggest a few alternative sampling strategies and compare them with recently proposed methods.

Uncertainty principle and sampling of signals defined on graphs

In many applications, from sensor to social networks, gene regulatory networks or big data, observations can be represented as a signal defined over the vertices of a graph. Building on the recently introduced Graph Fourier Transform, the first contribution of this paper is to provide an uncertainty principle for signals on graph. As a by-product of this theory, we show how to build a dictionary of maximally concentrated signals on vertex/frequency domains. Then, we establish a direct relation between uncertainty principle and sampling, which forms the basis for a sampling theorem of signals defined on graph. Based on this theory, we show that, besides sampling rate, the samples' location plays a key role in the performance of signal recovery algorithms. Hence, we suggest a few alternative sampling strategies and compare them with recently proposed methods.

Mikhail Tsitsvero

Papers

Signals on Graphs: Uncertainty Principle and Sampling

An introduction to hypergraph signal processing

On the degrees of freedom of signals on graphs

Uncertainty principle and sampling of signals defined on graphs

Uncertainty Principle and Sampling of Signals Defined on Graphs