Variational models for signal processing with graph neural networks

DOI10.1007/978-3-030-75549-2_26MaRDI QIDQ826200zbMATH OpenOpenAlexFDO

Authors A. Azad, Julien Rabin, Abderrahim Elmoataz

Publication date 20 December 2021

Full work available at URL https://arxiv.org/abs/2103.16337

zbMATH Keywords

total variationunsupervised learningneural networkvariational methodsgraph processingmessage-passing network

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Artificial neural networks and deep learning (68T07) Signal theory (characterization, reconstruction, filtering, etc.) (94A12) Variational inequalities (49J40)

Abstract: This paper is devoted to signal processing on point-clouds by means of neural networks. Nowadays, state-of-the-art in image processing and computer vision is mostly based on training deep convolutional neural networks on large datasets. While it is also the case for the processing of point-clouds with Graph Neural Networks (GNN), the focus has been largely given to high-level tasks such as classification and segmentation using supervised learning on labeled datasets such as ShapeNet. Yet, such datasets are scarce and time-consuming to build depending on the target application. In this work, we investigate the use of variational models for such GNN to process signals on graphs for unsupervised learning. Our contributions are two-fold. We first show that some existing variational-based algorithms for signals on graphs can be formulated as Message Passing Networks (MPN), a particular instance of GNN, making them computationally efficient in practice when compared to standard gradient-based machine learning algorithms. Secondly, we investigate the unsupervised learning of feed-forward GNN, either by direct optimization of an inverse problem or by model distillation from variational-based MPN. Keywords:Graph Processing. Neural Network. Total Variation. Variational Methods. Message Passing Network. Unsupervised learning

Recommendations

• Graph convolutional neural networks via scattering
• Fast Haar transforms for graph neural networks
• Understanding the message passing in graph neural networks via power iteration clustering
• Convex variational methods on graphs for multiclass segmentation of high-dimensional data and point clouds
• Pseudoinverse graph convolutional networks. Fast filters tailored for large eigengaps of dense graphs and hypergraphs

Cites work

• A first-order primal-dual algorithm for convex problems with applications to imaging
• Deep neural network structures solving variational inequalities
• Deep unfolding of a proximal interior point method for image restoration
• Non-local regularization of inverse problems
• Nonlinear multilayered representation of graph-signals
• Nonlocal Operators with Applications to Image Processing
• On the \(p\)-Laplacian and \(\infty\)-Laplacian on graphs with applications in image and data processing
• On the limited memory BFGS method for large scale optimization
• Parseval proximal neural networks
• Preconditioning of a generalized forward-backward splitting and application to optimization on graphs

Cited in

(2)

• Revisiting graph neural networks from hybrid regularized graph signal reconstruction
• Understanding the message passing in graph neural networks via power iteration clustering

This page was built for publication: Variational models for signal processing with graph neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q826200)