{"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/combining-label-propagation-and-simple-models-1","title":"Combining Label Propagation and Simple Models Out-performs Graph Neural Networks","arxiv_id":"2010.13993","date":"2020-10-27","proceeding":"ICLR 2021 1","authors":["Qian Huang","Horace He","Abhay Singh","Ser-Nam Lim","Austin R. Benson"],"abstract":"Graph Neural Networks (GNNs) are the predominant technique for learning over graphs. However, there is relatively little understanding of why GNNs are successful in practice and whether they are necessary for good performance. Here, we show that for many standard transductive node classification benchmarks, we can exceed or match the performance of state-of-the-art GNNs by combining shallow models that ignore the graph structure with two simple post-processing steps that exploit correlation in the label structure: (i) an \"error correlation\" that spreads residual errors in training data to correct errors in test data and (ii) a \"prediction correlation\" that smooths the predictions on the test data. We call this overall procedure Correct and Smooth (C&S), and the post-processing steps are implemented via simple modifications to standard label propagation techniques from early graph-based semi-supervised learning methods. Our approach exceeds or nearly matches the performance of state-of-the-art GNNs on a wide variety of benchmarks, with just a small fraction of the parameters and orders of magnitude faster runtime. For instance, we exceed the best known GNN performance on the OGB-Products dataset with 137 times fewer parameters and greater than 100 times less training time. The performance of our methods highlights how directly incorporating label information into the learning algorithm (as was done in traditional techniques) yields easy and substantial performance gains. We can also incorporate our techniques into big GNN models, providing modest gains. Our code for the OGB results is at https://github.com/Chillee/CorrectAndSmooth.","url_abs":"https://arxiv.org/abs/2010.13993v2","url_pdf":"https://arxiv.org/pdf/2010.13993v2.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":"combining-label-propagation-and-simple-models-1","repo_url":"https://github.com/CUAI/CorrectAndSmooth","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"combining-label-propagation-and-simple-models-1","repo_url":"https://github.com/Chillee/CorrectAndSmoothOGB","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"combining-label-propagation-and-simple-models-1","repo_url":"https://github.com/sangyx/gtrick/tree/main/benchmark/pyg","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"combining-label-propagation-and-simple-models-1","repo_url":"https://github.com/xnuohz/CorrectAndSmooth-dgl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"combining-label-propagation-and-simple-models-1","repo_url":"https://github.com/ytchx1999/GCN_res-CS-v2","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"combining-label-propagation-and-simple-models-1","repo_url":"https://github.com/ytchx1999/PyG-GCN_res-CS","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"combining-label-propagation-and-simple-models-1","repo_url":"https://github.com/dmlc/dgl/tree/master/examples/pytorch/correct_and_smooth","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"node-classification","task_name":"Node Classification"},{"task_slug":"node-classification-on-non-homophilic","task_name":"Node Classification on Non-Homophilic (Heterophilic) Graphs"},{"task_slug":"node-property-prediction","task_name":"Node Property Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/node-classification-on-penn94","task":"Node Classification","dataset":"Penn94","model":"C&S 2-hop","rank_in_archive_order":24,"of":32,"metrics":{"Accuracy":"78.40 ± 3.12"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-penn94","task":"Node Classification","dataset":"Penn94","model":"C&S 1-hop","rank_in_archive_order":28,"of":32,"metrics":{"Accuracy":"74.28 ± 1.19"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-genius","task":"Node Classification","dataset":"genius","model":"C&S 2-hop","rank_in_archive_order":17,"of":26,"metrics":{"Accuracy":"84.94 ± 0.49"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-genius","task":"Node Classification","dataset":"genius","model":"C&S 1-hop","rank_in_archive_order":18,"of":26,"metrics":{"Accuracy":"82.93 ± 0.15"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-6","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"Deezer-Europe","model":"C&S(1hop)","rank_in_archive_order":19,"of":28,"metrics":{"1:1 Accuracy":"64.60±0.57"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-6","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"Deezer-Europe","model":"C&S(2hop)","rank_in_archive_order":20,"of":28,"metrics":{"1:1 Accuracy":"64.52±0.62"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-13","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"Penn94","model":"C&S 2-hop","rank_in_archive_order":20,"of":28,"metrics":{"1:1 Accuracy":"78.40 ± 3.12"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-13","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"Penn94","model":"C&S 1-hop","rank_in_archive_order":24,"of":28,"metrics":{"1:1 Accuracy":"74.28 ± 1.19"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-14","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"genius","model":"C&S 2-hop","rank_in_archive_order":19,"of":26,"metrics":{"1:1 Accuracy":"84.94 ± 0.49"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-14","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"genius","model":"C&S 1-hop","rank_in_archive_order":20,"of":26,"metrics":{"1:1 Accuracy":"82.93 ± 0.15"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-15","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"twitch-gamers","model":"C&S 2-hop","rank_in_archive_order":11,"of":26,"metrics":{"1:1 Accuracy":"65.02 ± 0.16"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-non-homophilic-15","task":"Node Classification on Non-Homophilic (Heterophilic) Graphs","dataset":"twitch-gamers","model":"C&S 1-hop","rank_in_archive_order":12,"of":26,"metrics":{"1:1 Accuracy":"64.86 ± 0.27"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-arxiv","task":"Node Property Prediction","dataset":"ogbn-arxiv","model":"GAT(norm.adj.)+label reuse+C&S","rank_in_archive_order":33,"of":86,"metrics":{"Ext. data":"No","Number of params":"1441580","Test Accuracy":"0.7395 ± 0.0012","Validation Accuracy":"0.7519 ± 0.0008"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-arxiv","task":"Node Property Prediction","dataset":"ogbn-arxiv","model":"GAT + C&S","rank_in_archive_order":35,"of":86,"metrics":{"Ext. data":"No","Number of params":"1567000","Test Accuracy":"0.7386 ± 0.0014","Validation Accuracy":"0.7484 ± 0.0007"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-arxiv","task":"Node Property Prediction","dataset":"ogbn-arxiv","model":"GCN_res + C&S_v2","rank_in_archive_order":46,"of":86,"metrics":{"Ext. data":"No","Number of params":"155824","Test Accuracy":"0.7313 ± 0.0017","Validation Accuracy":"0.7445 ± 0.0011"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-arxiv","task":"Node Property Prediction","dataset":"ogbn-arxiv","model":"MLP + C&S","rank_in_archive_order":47,"of":86,"metrics":{"Ext. data":"No","Number of params":"175656","Test Accuracy":"0.7312 ± 0.0012","Validation Accuracy":"0.7391 ± 0.0015"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-arxiv","task":"Node Property Prediction","dataset":"ogbn-arxiv","model":"GCN_res + C&S","rank_in_archive_order":51,"of":86,"metrics":{"Ext. data":"No","Number of params":"155824","Test Accuracy":"0.7297 ± 0.0022","Validation Accuracy":"0.7423 ± 0.0014"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-arxiv","task":"Node Property Prediction","dataset":"ogbn-arxiv","model":"Linear + C&S","rank_in_archive_order":63,"of":86,"metrics":{"Ext. data":"No","Number of params":"15400","Test Accuracy":"0.7222 ± 0.0002","Validation Accuracy":"0.7368 ± 0.0004"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-arxiv","task":"Node Property Prediction","dataset":"ogbn-arxiv","model":"Plain Linear + C&S","rank_in_archive_order":78,"of":86,"metrics":{"Ext. data":"No","Number of params":"5160","Test Accuracy":"0.7126 ± 0.0001","Validation Accuracy":"0.7300 ± 0.0001"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"Spec-MLP-Wide + C&S","rank_in_archive_order":21,"of":64,"metrics":{"Ext. data":"No","Number of params":"406063","Test Accuracy":"0.8451 ± 0.0006","Validation Accuracy":"0.9132 ± 0.0010"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"MLP + C&S","rank_in_archive_order":24,"of":64,"metrics":{"Ext. data":"No","Number of params":"96247","Test Accuracy":"0.8418 ± 0.0007","Validation Accuracy":"0.9147 ± 0.0009"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"Linear + C&S","rank_in_archive_order":31,"of":64,"metrics":{"Ext. data":"No","Number of params":"10763","Test Accuracy":"0.8301 ± 0.0001","Validation Accuracy":"0.9134 ± 0.0001"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"Plain Linear + C&S","rank_in_archive_order":33,"of":64,"metrics":{"Ext. data":"No","Number of params":"4747","Test Accuracy":"0.8254 ± 0.0003","Validation Accuracy":"0.9103 ± 0.0001"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GAT w/NS + C&S","rank_in_archive_order":42,"of":64,"metrics":{"Ext. data":"No","Number of params":"753622","Test Accuracy":"0.8092 ± 0.0037","Validation Accuracy":"0.9263 ± 0.0008"},"uses_additional_data":false},{"leaderboard":"/sota/node-property-prediction-on-ogbn-products","task":"Node Property Prediction","dataset":"ogbn-products","model":"GraphSAGE w/NS + C&S","rank_in_archive_order":45,"of":64,"metrics":{"Ext. data":"No","Number of params":"207919","Test Accuracy":"0.8041 ± 0.0022","Validation Accuracy":"0.9238 ± 0.0007"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2010.13993","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2010.13993"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ytchx1999/GCN_res-CS-v2","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Chillee/CorrectAndSmoothOGB","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ytchx1999/PyG-GCN_res-CS","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/xnuohz/CorrectAndSmooth-dgl","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/sangyx/gtrick/tree/main/benchmark/pyg","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/CUAI/CorrectAndSmooth","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/dmlc/dgl/tree/master/examples/pytorch/correct_and_smooth","reach":null}],"summary":{"ran_draft_wrong":2,"unverified":3},"by_repo_kind":{"official":{"samples":2,"ran":2,"repositories":1},"listed":{"samples":3,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":2,"samples":[{"code_sha256_prefix":"5a209cddf7b3e52f","entry":"test","repo":"CUAI/CorrectAndSmooth","repo_kind":"official","path":"gen_models.py","file_url":"https://github.com/CUAI/CorrectAndSmooth/blob/HEAD/gen_models.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"5a209cddf7b3e52f"}},{"code_sha256_prefix":"fac9425843013f50","entry":"train","repo":"CUAI/CorrectAndSmooth","repo_kind":"official","path":"gen_models.py","file_url":"https://github.com/CUAI/CorrectAndSmooth/blob/HEAD/gen_models.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"fac9425843013f50"}},{"code_sha256_prefix":"3f05a96b5815dd82","entry":"eval","repo":"sangyx/gtrick","repo_kind":"listed","path":"benchmark/dgl/graph_pred.py","file_url":"https://github.com/sangyx/gtrick/blob/HEAD/benchmark/dgl/graph_pred.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"3f05a96b5815dd82"}},{"code_sha256_prefix":"95753adaeccb24eb","entry":"random_feature","repo":"sangyx/gtrick","repo_kind":"listed","path":"gtrick/random_feature.py","file_url":"https://github.com/sangyx/gtrick/blob/HEAD/gtrick/random_feature.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"95753adaeccb24eb"}},{"code_sha256_prefix":"4254dbfb6411b68f","entry":"train","repo":"sangyx/gtrick","repo_kind":"listed","path":"benchmark/dgl/graph_pred.py","file_url":"https://github.com/sangyx/gtrick/blob/HEAD/benchmark/dgl/graph_pred.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4254dbfb6411b68f"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}