Deep Convolutional Networks on Graph-Structured Data

2015/06/16 by Mikael Henaff, Joan Bruna, Henaff, Mikael +3 · 77 citations
Computer Science · #Advanced Graph Neural Networks #Advanced Neural Network Applications #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neural and Evolutionary Computing (cs.NE)

paper · pdf · doi:10.48550/arxiv.1506.05163

openalex publication_date 2015/06/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Deep Learning's recent successes have mostly relied on Convolutional Networks, which exploit fundamental statistical properties of images, sounds and video data: the local stationarity and multi-scale compositional structure, that allows expressing long range interactions in terms of shorter, localized interactions. However, there exist other important examples, such as text documents or bioinformatic data, that may lack some or all of these strong statistical regularities. In this paper we consider the general question of how to construct deep architectures with small learning complexity on general non-Euclidean domains, which are typically unknown and need to be estimated from the data. In particular, we develop an extension of Spectral Networks which incorporates a Graph Estimation procedure, that we test on large-scale classification problems, matching or improving over Dropout Networks with far less parameters to estimate.

Cited by

• Personalized Pricing in Social Networks with Individual and Group Fairness Considerations
• Edged Weisfeiler-Lehman Algorithm
• Mitigating Structural Overfitting: A Distribution-Aware Rectification Framework for Missing Feature Imputation
• Efficient Curvature-aware Graph Network
• How Wide and How Deep? Mitigating Over-Squashing of GNNs via Channel Capacity Constrained Estimation
• Understanding Graph Neural Networks from Graph Signal Denoising Perspectives
• Curvature Graph Neural Network
• Learning Feature Aggregation for Deep 3D Morphable Models
• Self-Attention Graph Pooling
• Local Propagation in Constraint-based Neural Network
• Image augmentation with invertible networks in interactive satellite image change detection
• Distributed Cartesian Power Graph Segmentation for Graphon Estimation
• Noise-robust classification with hypergraph neural network
• Graph Neural Networks: A Review of Methods and Applications
• Machine Learning on Graphs: A Model and Comprehensive Taxonomy
• Say No to the Discrimination: Learning Fair Graph Neural Networks with Limited Sensitive Attribute Information
• SportsCap: Monocular 3D Human Motion Capture and Fine-grained Understanding in Challenging Sports Videos
• Convolutional Imputation of Matrix Networks
• Relational Graph Attention Networks
• Hierarchical Graph-to-Graph Translation for Molecules
• Label Efficient Semi-Supervised Learning via Graph Filtering
• DropEdge: Towards Deep Graph Convolutional Networks on Node Classification
• Spatio-Temporal Inception Graph Convolutional Networks for Skeleton-Based Action Recognition
• Label-frugal satellite image change detection with generative virtual exemplar learning
• Deep Learning and Quantum Entanglement: Fundamental Connections with\n Implications to Network Design
• FedGL: Federated Graph Learning Framework with Global Self-Supervision
• A Pixel-Based Framework for Data-Driven Clothing
• A Graph-CNN for 3D Point Cloud Classification
• Graph networks as learnable physics engines for inference and control
• MeshCNN
• Iterative Visual Reasoning Beyond Convolutions
• Disentangling and Unifying Graph Convolutions for Skeleton-Based Action Recognition
• Topological based classification using graph convolutional networks
• Quantum Graph Neural Networks
• ReLaText: Exploiting Visual Relationships for Arbitrary-Shaped Scene Text Detection with Graph Convolutional Networks
• Adaptive Graph Convolutional Network with Attention Graph Clustering for Co-saliency Detection
• Analyzing the Performance of Graph Neural Networks with Pipe Parallelism
• Unsupervised Learning of Graph Hierarchical Abstractions with Differentiable Coarsening and Optimal Transport
• Tensor Programs IIb: Architectural Universality of Neural Tangent Kernel Training Dynamics
• MILE: A Multi-Level Framework for Scalable Graph Embedding
• Mesh Variational Autoencoders with Edge Contraction Pooling
• Graph Denoising with Framelet Regularizer
• MeshWalker: Deep Mesh Understanding by Random Walks
• Junction Tree Variational Autoencoder for Molecular Graph Generation
• G3CN: Gaussian Topology Refinement Gated Graph Convolutional Network for Skeleton-Based Action Recognition
• PasteGAN: A Semi-Parametric Method to Generate Image from Scene Graph
• Semi-supervised Learning with Adaptive Neighborhood Graph Propagation Network
• LinkNet: Relational Embedding for Scene Graph
• Neural Graph Evolution: Towards Efficient Automatic Robot Design
• Graph Convolutional Networks for Text Classification
• Pedestrian Attribute Recognition: A Survey
• SyncSpecCNN: Synchronized Spectral CNN for 3D Shape Segmentation
• Deep Feature Mining via Attention-based BiLSTM-GCN for Human Motor Imagery Recognition
• Bayesian Graph Convolutional Neural Networks using Node Copying
• Constructing Geographic and Long-term Temporal Graph for Traffic Forecasting
• DeeperGCN: All You Need to Train Deeper GCNs
• Unifying local and non-local signal processing with graph CNNs
• Learning the Implicit Semantic Representation on Graph-Structured Data
• Saliency Prediction with External Knowledge
• Human Action Recognition with Multi-Laplacian Graph Convolutional Networks
• Fast Haar Transforms for Graph Neural Networks
• A deep learning approach to real-time parking occupancy prediction in spatio-temporal networks incorporating multiple spatio-temporal data sources
• ManifoldNet: A Deep Network Framework for Manifold-valued Data
• Graph Capsule Convolutional Neural Networks
• Gromov-Wasserstein Factorization Models for Graph Clustering
• DGCNN: Disordered Graph Convolutional Neural Network Based on the Gaussian Mixture Model
• RGCNN: Regularized Graph CNN for Point Cloud Segmentation
• Rubik: A Hierarchical Architecture for Efficient Graph Learning
• RigNet: Neural Rigging for Articulated Characters
• Neighbor-Sampling Based Momentum Stochastic Methods for Training Graph Neural Networks
• Lookup subnet based Spatial Graph Convolutional neural Network
• FPConv: Learning Local Flattening for Point Convolution
• A Comprehensive Survey of Graph Embedding: Problems, Techniques and Applications
• Higher-order Graph Convolutional Networks
• Learning to Infer Unseen Single-/Multi-Attribute-Object Compositions with Graph Networks
• Probabilistic modeling of lake surface water temperature using a Bayesian spatio-temporal graph convolutional neural network
• CommPOOL: An Interpretable Graph Pooling Framework for Hierarchical Graph Representation Learning

Related

• Spectral Networks and Locally Connected Networks on Graphs
• Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks
• A Generalization of Convolutional Neural Networks to Graph-Structured Data
• A Convolutional Neural Network into graph space
• Bridging the Gap between Spatial and Spectral Domains: A Survey on Graph Neural Networks