Revisiting Semi-Supervised Learning with Graph Embeddings

Abstract

We present a semi-supervised learning framework based on graph embeddings. Given a graph between instances, we train an embedding for each instance to jointly predict the class label and the neighborhood context in the graph. We develop both transductive and inductive variants of our method. In the transductive variant of our method, the class labels are determined by both the learned embeddings and input feature vectors, while in the inductive variant, the embeddings are defined as a parametric function of the feature vectors, so predictions can be made on instances not seen during training. On a large and diverse set of benchmark tasks, including text classification, distantly supervised entity extraction, and entity classification, we show improved performance over many of the existing models.

ICML 2016

References in corpus (2)

• Natural Language Processing (almost) from Scratch
• Multi-Task Cross-Lingual Sequence Tagging from Scratch

Cited by in corpus (63)

• Unsupervised Data Augmentation for Consistency Training
• Adversarial Attack and Defense on Graph Data: A Survey
• Self-training with Noisy Student improves ImageNet classification
• Good Semi-supervised Learning that Requires a Bad GAN
• Semi-Supervised Deep Learning for Fully Convolutional Networks
• GRAM: Graph-based Attention Model for Healthcare Representation Learning
• Towards Deeper Graph Neural Networks with Differentiable Group Normalization
• Deep Models of Interactions Across Sets
• Revisiting Over-smoothing in Deep GCNs
• Graph Partition Neural Networks for Semi-Supervised Classification
• A Tutorial on Network Embeddings
• Dirichlet Energy Constrained Learning for Deep Graph Neural Networks
• A Flexible Generative Framework for Graph-based Semi-supervised Learning
• Grale: Designing Networks for Graph Learning
• Graphite: Iterative Generative Modeling of Graphs
• CrossATNet - A Novel Cross-Attention Based Framework for Sketch-Based Image Retrieval
• Edge-Featured Graph Attention Network
• Domain-adaptive Message Passing Graph Neural Network
• Active Learning for Graph Neural Networks via Node Feature Propagation
• Graph Policy Network for Transferable Active Learning on Graphs
• Bayesian Semi-supervised Learning with Graph Gaussian Processes
• PAN: Path Integral Based Convolution for Deep Graph Neural Networks
• Learning to Drop: Robust Graph Neural Network via Topological Denoising
• GraphZoom: A multi-level spectral approach for accurate and scalable graph embedding
• Bayesian Graph Convolutional Neural Networks using Node Copying
• Generative Graph Convolutional Network for Growing Graphs
• Deep Co-Space: Sample Mining Across Feature Transformation for Semi-Supervised Learning
• Latent Variable Modelling with Hyperbolic Normalizing Flows
• PPGN: Physics-Preserved Graph Networks for Real-Time Fault Location in Distribution Systems with Limited Observation and Labels
• Adaptively Connected Neural Networks
• Weakly Supervised Learning for Analyzing Political Campaigns on Facebook
• Infinitely Wide Graph Convolutional Networks: Semi-supervised Learning via Gaussian Processes
• Identifying Key Nodes for the Influence Spread using a Machine Learning Approach
• The Impact of Global Structural Information in Graph Neural Networks Applications
• Self-Enhanced GNN: Improving Graph Neural Networks Using Model Outputs
• Graph-Revised Convolutional Network
• SPINE: Structural Identity Preserved Inductive Network Embedding
• Training Matters: Unlocking Potentials of Deeper Graph Convolutional Neural Networks
• ANAE: Learning Node Context Representation for Attributed Network Embedding
• Topological based classification using graph convolutional networks
• Topology and Content Co-Alignment Graph Convolutional Learning
• Joint Learning of Graph Representation and Node Features in Graph Convolutional Neural Networks
• Towards Train-Test Consistency for Semi-supervised Temporal Action Localization
• Semi-Supervised Learning with Declaratively Specified Entropy Constraints
• Exploring Structure-Adaptive Graph Learning for Robust Semi-Supervised Classification
• Topological based classification of paper domains using graph convolutional networks
• Collaborative Graph Walk for Semi-supervised Multi-Label Node Classification
• Progressive Representative Labeling for Deep Semi-Supervised Learning
• Embedding Graphs on Grassmann Manifold
• Learning the Implicit Semantic Representation on Graph-Structured Data
• Recurrent Attention Walk for Semi-supervised Classification
• Enhancing Network Embedding with Auxiliary Information: An Explicit Matrix Factorization Perspective
• Towards an Efficient and General Framework of Robust Training for Graph Neural Networks
• Models for information propagation on graphs
• Node Embedding via Word Embedding for Network Community Discovery
• Bayesian Attention Modules
• Exploiting Transductive Property of Graph Convolutional Neural Networks with Less Labeling Effort
• Enhanced Network Embeddings via Exploiting Edge Labels
• Alleviating Over-Smoothing via Aggregation over Compact Manifolds
• Semi-Supervised Deep Learning Using Improved Unsupervised Discriminant Projection
• Policy Message Passing: A New Algorithm for Probabilistic Graph Inference
• Snowball: Iterative Model Evolution and Confident Sample Discovery for Semi-Supervised Learning on Very Small Labeled Datasets
• Counterfactual Propagation for Semi-Supervised Individual Treatment Effect Estimation