REVIEW 7 cited by
InfoGraph: Unsupervised and Semi-supervised Graph-Level Representation Learning via Mutual Information Maximization
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
This paper studies learning the representations of whole graphs in both unsupervised and semi-supervised scenarios. Graph-level representations are critical in a variety of real-world applications such as predicting the properties of molecules and community analysis in social networks. Traditional graph kernel based methods are simple, yet effective for obtaining fixed-length representations for graphs but they suffer from poor generalization due to hand-crafted designs. There are also some recent methods based on language models (e.g. graph2vec) but they tend to only consider certain substructures (e.g. subtrees) as graph representatives. Inspired by recent progress of unsupervised representation learning, in this paper we proposed a novel method called InfoGraph for learning graph-level representations. We maximize the mutual information between the graph-level representation and the representations of substructures of different scales (e.g., nodes, edges, triangles). By doing so, the graph-level representations encode aspects of the data that are shared across different scales of substructures. Furthermore, we further propose InfoGraph*, an extension of InfoGraph for semi-supervised scenarios. InfoGraph* maximizes the mutual information between unsupervised graph representations learned by InfoGraph and the representations learned by existing supervised methods. As a result, the supervised encoder learns from unlabeled data while preserving the latent semantic space favored by the current supervised task. Experimental results on the tasks of graph classification and molecular property prediction show that InfoGraph is superior to state-of-the-art baselines and InfoGraph* can achieve performance competitive with state-of-the-art semi-supervised models.
Forward citations
Cited by 7 Pith papers
-
Frequency-Corrupt Based Graph Self-Supervised Learning
FC-GSSL is a graph self-supervised method that corrupts nodes/edges with high low-frequency contribution and reconstructs low-frequency/general targets, improving node and graph prediction on most tested benchmarks.
-
Graph Positional Autoencoders as Self-supervised Learners
A dual-path graph autoencoder that reconstructs node features and Laplacian-eigenvector distances reports strong self-supervised results on heterophilic and molecular benchmarks, with some overstatement in the margins...
-
Learnable Kernel Density Estimation for Graphs and Its Application to Graph-Level Anomaly Detection
LGKDE learns a maximum mean discrepancy based graph metric and fits a multi-scale kernel density estimator, using perturbed graphs as contrastive targets for graph-level anomaly detection.
-
Khan-GCL: Kolmogorov-Arnold Network Based Graph Contrastive Learning with Hard Negatives
Khan-GCL combines KAN encoders with coefficient-based critical feature identification to generate hard negatives and reports state-of-the-art graph classification results.
-
DICE: Device-level Integrated Circuits Encoder with Graph Contrastive Pretraining
A self-supervised graph encoder for device-level circuits improves prediction of circuit similarity, delays, and op-amp performance across analog and digital designs.
-
Closer through commonality: Enhancing hypergraph contrastive learning with shared groups
HyFi improves hypergraph contrastive learning by weighting nodes that share hyperedges as 'weak positives' and using feature noise instead of topology corruption for augmentation.
-
Scalable Attribute-Missing Graph Clustering via Neighborhood Differentiation
CMV-ND builds one feature view per exact hop distance and then runs existing clustering methods on those views, improving attribute-missing large-scale graph clustering in experiments.
Discussion (0). Continue with ORCID to comment.