InfoGCL: Information-Aware Graph Contrastive Learning

Authors: Dongkuan Xu, Wei Cheng, Dongsheng Luo, Haifeng Chen, Xiang Zhang

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically validate our theoretical analysis on both node and graph classification benchmark datasets, and demonstrate that our algorithm significantly outperforms the state-of-the-arts.
Researcher Affiliation Collaboration 1The Pennsylvania State University 2NEC Labs America 1{dux19,dul262,xzz89}@psu.edu 2{weicheng,haifeng}@nec-labs.com
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to source code for the methodology described in this paper.
Open Datasets Yes We use both graph classification and node classification benchmark datasets that are widely used in the existing graph contrastive learning approaches. The graph classification datasets include MUTAG [17], PTC-MR [17], IMDB-B [40], IMDB-M [40], NCI1 [34], and COLLAB [40]. The node classification datasets include Citeseer, Cora, and Pubmed [23].
Dataset Splits Yes For graph classification, we report the mean 10-fold cross validation accuracy after 5 runs followed by a linear SVM. The linear SVM is trained by applying cross validation on training data folds and the best mean accuracy is reported.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup Yes We conduct experiment with the values of the number of GNN layers, the number of epochs, batch size, the parameter C of SVM in the sets {2, 4, 8, 12}, {10, 20, 40, 100}, {32, 64, 128, 256} and {10 3, 10 2, ..., 102, 103 }, respectively. We adopt the basic setting of DGI for node classification. Specifically, we set the number of GNN layers to 1 and experiment with the batch size in the set {2, 4, 8}. The hidden dimension of representations is set to 512. We also apply the early stopping strategy.