{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/graphsaint-graph-sampling-based-inductive","title":"GraphSAINT: Graph Sampling Based Inductive Learning Method","arxiv_id":"1907.04931","date":"2019-07-10","proceeding":"ICLR 2020 1","authors":["Hanqing Zeng","Hongkuan Zhou","Ajitesh Srivastava","Rajgopal Kannan","Viktor Prasanna"],"abstract":"Graph Convolutional Networks (GCNs) are powerful models for learning representations of attributed graphs. To scale GCNs to large graphs, state-of-the-art methods use various layer sampling techniques to alleviate the \"neighbor explosion\" problem during minibatch training. We propose GraphSAINT, a graph sampling based inductive learning method that improves training efficiency and accuracy in a fundamentally different way. By changing perspective, GraphSAINT constructs minibatches by sampling the training graph, rather than the nodes or edges across GCN layers. Each iteration, a complete GCN is built from the properly sampled subgraph. Thus, we ensure fixed number of well-connected nodes in all layers. We further propose normalization technique to eliminate bias, and sampling algorithms for variance reduction. Importantly, we can decouple the sampling from the forward and backward propagation, and extend GraphSAINT with many architecture variants (e.g., graph attention, jumping connection). GraphSAINT demonstrates superior performance in both accuracy and training time on five large graphs, and achieves new state-of-the-art F1 scores for PPI (0.995) and Reddit (0.970).","url_abs":"https://arxiv.org/abs/1907.04931v4","url_pdf":"https://arxiv.org/pdf/1907.04931v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/GraphSAINT/GraphSAINT","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null},{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/GraphSAINT/GraphACT","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/hyeamykim/GCN-related-works","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/lt610/GraphSaint","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/maysambehmanesh/SGCL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/thudm/graphmae2","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/xingsumq/us-defake","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"graphsaint-graph-sampling-based-inductive","repo_url":"https://github.com/dmlc/dgl/tree/master/examples/pytorch/graphsaint","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"graph-attention","task_name":"Graph Attention"},{"task_slug":"graph-embedding","task_name":"Graph Embedding"},{"task_slug":"graph-representation-learning","task_name":"Graph Representation Learning"},{"task_slug":"graph-sampling","task_name":"Graph Sampling"},{"task_slug":"inductive-learning","task_name":"Inductive Learning"},{"task_slug":"node-classification","task_name":"Node Classification"},{"task_slug":"node-property-prediction","task_name":"Node Property Prediction"}],"methods":[{"method_slug":"gcn","method_name":"GCN"},{"method_slug":"graphsaint","method_name":"GraphSAINT"}],"datasets_introduced":[],"methods_introduced":[{"slug":"graphsaint","name":"GraphSAINT","full_name":"Graph sampling based inductive learning method"}],"results":[{"leaderboard":"/sota/link-property-prediction-on-ogbl-citation2","task":"Link Property Prediction","dataset":"ogbl-citation2","model":"GraphSAINT (GCN aggr)","rank_in_archive_order":19,"of":23,"metrics":{"Ext. data":"No","Number of params":"296449","Test MRR":"0.7985 ± 0.0040","Validation MRR":"0.7975 ± 0.0039"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-ppi","task":"Node Classification","dataset":"PPI","model":"GraphSAINT","rank_in_archive_order":3,"of":24,"metrics":{"F1":"99.50"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-reddit","task":"Node Classification","dataset":"Reddit","model":"GraphSAINT","rank_in_archive_order":6,"of":16,"metrics":{"Accuracy":"97.0%"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-mag","task":"Node Property Prediction","dataset":"ogbn-mag","model":"GraphSAINT + metapath2vec","rank_in_archive_order":28,"of":39,"metrics":{"Ext. data":"No","Number of params":"309764724","Test Accuracy":"0.4966 ± 0.0022","Validation Accuracy":"0.5066 ± 0.0017"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-mag","task":"Node Property Prediction","dataset":"ogbn-mag","model":"GraphSAINT (R-GCN aggr)","rank_in_archive_order":30,"of":39,"metrics":{"Ext. data":"No","Number of params":"154366772","Test Accuracy":"0.4751 ± 0.0022","Validation Accuracy":"0.4837 ± 0.0026"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GraphSAINT-inductive","rank_in_archive_order":46,"of":64,"metrics":{"Ext. data":"No","Number of params":"331661","Test Accuracy":"0.8027 ± 0.0026","Validation Accuracy":"Please tell us"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GraphSAINT (SAGE aggr)","rank_in_archive_order":51,"of":64,"metrics":{"Ext. data":"No","Number of params":"206895","Test Accuracy":"0.7908 ± 0.0024","Validation Accuracy":"0.9162 ± 0.0008"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1907.04931","atlas_url":"https://app.syntology.ai/?focus=1907.04931","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}