Abstract:Graph classification aims to predict the property of the whole graph, which has attracted growing attention in the graph learning community. This problem has been extensively studied in the literature of both graph convolutional networks and graph kernels. Graph convolutional networks can learn effective node representations via message passing to mine graph topology in an implicit way, whereas graph kernels can explicitly utilize graph structural knowledge for classification. Due to the scarcity of labeled data in real-world applications, semi-supervised algorithms are anticipated for this problem. In this paper, we propose Graph Harmonic Neural Network (GHNN) which combines the advantages of both worlds to sufficiently leverage the unlabeled data, and thus overcomes label scarcity in semi-supervised scenarios. Specifically, our GHNN consists of a graph convolutional network (GCN) module and a graph kernel network (GKN) module that explore graph topology information from complementary perspectives. To fully leverage the unlabeled data, we develop a novel harmonic contrastive loss and a harmonic consistency loss to harmonize the training of two modules by giving priority to high-quality unlabeled data, thereby reconciling prediction consistency between both of them. In this manner, the two modules mutually enhance each other to sufficiently explore the graph topology of both labeled and unlabeled data. Extensive experiments on a variety of benchmarks demonstrate the effectiveness of our approach over competitive baselines.

Modification of Semi-supervised Algorithm Based on Gaussian Random Fields and Harmonic Functions

Generalized entropy based semi-supervised learning

Improved Graph-based semi-supervised learning Schemes

A Semi-supervised Kernel Learning Method Based on Label Propagation

Semi-supervised Dimensionality Reduction Via Harmonic Functions

Spectral Transformation Approaches to Semi-supervised Learning

Semi-supervised Gaussian Random Field Transduction and Induction

A Regularized Approach for Geodesic-Based Semisupervised Multimanifold Learning

Semi-Supervised Hybrid Clustering by Integrating Gaussian Mixture Model and Distance Metric Learning

Harmonic Functions Based Semi-Supervised Learning for Web Spam Detection.

Adaptive Semi-Supervised Feature Selection for Cross-Modal Retrieval

Semi-Supervised Regression and Active Learning with GF

GHNN: Graph Harmonic Neural Networks for semi-supervised graph-level classification

Combining Clustering Coefficient-Based Active Learning and Semi-Supervised Learning on Networked Data

Semi-supervised Learning Using Nonnegative Matrix Factorization and Harmonic Functions

A Semi-Supervised Learning Algorithm Via Adaptive Laplacian Graph

Adaptive Hash Retrieval with Kernel Based Similarity.

Graph Based Multi-class Semi-supervised Learning Using Gaussian Process

Graph Based Semi-Supervised Learning Via Label Fitting

SemiGPC: Distribution-Aware Label Refinement for Imbalanced Semi-Supervised Learning Using Gaussian Processes

Further Advances on Bayesian Ying-Yang Harmony Learning