{"work":{"id":"cb4e089d-c63e-4231-9e6b-b2ccb82c5329","openalex_id":null,"doi":null,"arxiv_id":"1810.00826","raw_key":null,"title":"How Powerful are Graph Neural Networks?","authors":null,"authors_text":"Keyulu Xu, Weihua Hu, Jure Leskovec, Stefanie Jegelka","year":2018,"venue":"cs.LG","abstract":"Graph Neural Networks (GNNs) are an effective framework for representation learning of graphs. GNNs follow a neighborhood aggregation scheme, where the representation vector of a node is computed by recursively aggregating and transforming representation vectors of its neighboring nodes. Many GNN variants have been proposed and have achieved state-of-the-art results on both node and graph classification tasks. However, despite GNNs revolutionizing graph representation learning, there is limited understanding of their representational properties and limitations. Here, we present a theoretical framework for analyzing the expressive power of GNNs to capture different graph structures. Our results characterize the discriminative power of popular GNN variants, such as Graph Convolutional Networks and GraphSAGE, and show that they cannot learn to distinguish certain simple graph structures. We then develop a simple architecture that is provably the most expressive among the class of GNNs and is as powerful as the Weisfeiler-Lehman graph isomorphism test. We empirically validate our theoretical findings on a number of graph classification benchmarks, and demonstrate that our model achieves state-of-the-art performance.","external_url":"https://arxiv.org/abs/1810.00826","cited_by_count":null,"metadata_source":"pith","metadata_fetched_at":"2026-05-25T04:55:23.410410+00:00","pith_arxiv_id":"1810.00826","created_at":"2026-05-09T23:24:42.992559+00:00","updated_at":"2026-05-25T04:55:23.410410+00:00","title_quality_ok":true,"display_title":"How Powerful are Graph Neural Networks?","render_title":"How Powerful are Graph Neural Networks?"},"hub":{"state":{"work_id":"cb4e089d-c63e-4231-9e6b-b2ccb82c5329","tier":"hub","tier_reason":"10+ Pith inbound or 1,000+ external citations","pith_inbound_count":69,"external_cited_by_count":null,"distinct_field_count":13,"first_pith_cited_at":"2019-09-03T17:10:28+00:00","last_pith_cited_at":"2026-05-22T10:01:28+00:00","author_build_status":"not_needed","summary_status":"needed","contexts_status":"needed","graph_status":"needed","ask_index_status":"not_needed","reader_status":"not_needed","recognition_status":"not_needed","updated_at":"2026-05-26T01:36:05.073739+00:00","tier_text":"hub"},"tier":"hub","role_counts":[{"context_role":"background","n":5},{"context_role":"method","n":3},{"context_role":"baseline","n":1}],"polarity_counts":[{"context_polarity":"background","n":5},{"context_polarity":"use_method","n":3},{"context_polarity":"baseline","n":1}],"runs":{"context_extract":{"job_type":"context_extract","status":"succeeded","result":{"enqueued_papers":25},"error":null,"updated_at":"2026-05-14T18:20:18.739641+00:00"},"graph_features":{"job_type":"graph_features","status":"succeeded","result":{"co_cited":[{"title":"Semi-Supervised Classification with Graph Convolutional Networks","work_id":"21fff118-807d-49cd-8229-f7087ba57b5d","shared_citers":18},{"title":"Graph Attention Networks","work_id":"7dd5bb04-b448-4f32-9719-5fc799641fd9","shared_citers":13},{"title":"ArXiv abs/2007.08663(2020)","work_id":"04426675-b2e3-4517-be09-08f2df683d14","shared_citers":4},{"title":"Inductive Representation Learning on Large Graphs","work_id":"a84007af-8130-4a77-b631-738c9af563fd","shared_citers":4},{"title":"Relational inductive biases, deep learning, and graph networks","work_id":"858410c0-7a66-4b27-b4e5-49aee9725be0","shared_citers":4},{"title":"Sdplib 1.2, a library of semidefinite program- ming test problems.Optimization Methods and Software, 11(1-4):683–690","work_id":"88e9cbb3-5e47-44fc-b1f0-f8bf82b9e07c","shared_citers":4},{"title":"arXiv preprint arXiv:1905.12265 , year=","work_id":"b6eca171-87f5-4e3b-8726-009d4d1b39d3","shared_citers":3},{"title":"Benchmarking graph neural networks","work_id":"87ffefbd-0e09-41bb-a247-fbc87d8d365e","shared_citers":3},{"title":"Neural message passing for quantum chemistry","work_id":"fb929706-486c-41f6-a1fd-96da310b2d62","shared_citers":3},{"title":"2010 , journal =","work_id":"1848424c-13f8-4544-a688-705f9b2422c2","shared_citers":2},{"title":"3d steerable cnns: Learning rotationally equivariant features in volumetric data.arXiv:1807.02547","work_id":"fbd44a4a-19a0-44c5-8f9e-43f5fc1d3ada","shared_citers":2},{"title":"Adam: A Method for Stochastic Optimization","work_id":"1910796d-9b52-4683-bf5c-de9632c1028b","shared_citers":2},{"title":"arXiv preprint arXiv:2203.06125 , year=","work_id":"f9af5e86-4da4-4d03-98e9-375c9dfb5dfa","shared_citers":2},{"title":"A Simple Structural Analysis Method for DAEs","work_id":"d8ce5cfc-20e8-456f-a19a-0b8698da903f","shared_citers":2},{"title":"Chemberta: Large-scale self- supervised pretraining for molecular property prediction","work_id":"44325e30-a45f-4052-8a05-3e9ed3f4c3cf","shared_citers":2},{"title":"Decoupled Weight Decay Regularization","work_id":"07ef7360-d385-4033-83f7-8384a6325204","shared_citers":2},{"title":"Fast Graph Representation Learning with PyTorch Geometric","work_id":"7030f6d8-1c95-4c18-a988-bfcf31d4cb4c","shared_citers":2},{"title":"Feinberg, Evan and Gomes, Joseph and Geniesse, Caleb and S","work_id":"6351d72d-6135-4484-afc8-abd85cfaa043","shared_citers":2},{"title":"How attentive are graph attention networks?arXiv preprint arXiv:2105.14491","work_id":"b81a77dc-03fc-40f2-a635-ba6c4a250268","shared_citers":2},{"title":"Invariant and equivariant graph networks.arXiv:1812.09902","work_id":"fb2e57cf-2565-45ea-8e6b-86753fbd52bf","shared_citers":2},{"title":"Kipf et al.Variational Graph Auto- Encoders","work_id":"2e8716eb-69bd-4617-8256-8543ba2c215c","shared_citers":2},{"title":"S., & Welling, M., 2016,Group Equivariant Convolu- tional Networks, Proc","work_id":"473fade8-b16b-45a0-b764-68a0f3957588","shared_citers":2},{"title":"Usbd: Universal structural basis distillation for source-free graph domain adaptation","work_id":"edabed69-11fa-4519-abcc-f26b69902a28","shared_citers":2},{"title":"Weidele, Claudio Bellei, Tom Robinson, and Charles E","work_id":"19e55ccd-457d-400e-b663-c70469ccb117","shared_citers":2}],"time_series":[{"n":34,"year":2026}],"dependency_candidates":[]},"error":null,"updated_at":"2026-05-14T18:26:23.270564+00:00"},"identity_refresh":{"job_type":"identity_refresh","status":"succeeded","result":{"items":[{"title":"Qwen3 Technical Report","outcome":"unchanged","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","resolver":"local_arxiv","confidence":0.98,"old_work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e"}],"counts":{"fixed":0,"merged":0,"unchanged":1,"quarantined":0,"needs_external_resolution":0},"errors":[],"attempted":1},"error":null,"updated_at":"2026-05-14T18:19:35.425023+00:00"},"summary_claims":{"job_type":"summary_claims","status":"succeeded","result":{"title":"How Powerful are Graph Neural Networks?","claims":[{"claim_text":"Graph Neural Networks (GNNs) are an effective framework for representation learning of graphs. GNNs follow a neighborhood aggregation scheme, where the representation vector of a node is computed by recursively aggregating and transforming representation vectors of its neighboring nodes. Many GNN variants have been proposed and have achieved state-of-the-art results on both node and graph classification tasks. However, despite GNNs revolutionizing graph representation learning, there is limited understanding of their representational properties and limitations. Here, we present a theoretical f","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks How Powerful are Graph Neural Networks? because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T18:26:19.338719+00:00"}},"summary":{"title":"How Powerful are Graph Neural Networks?","claims":[{"claim_text":"Graph Neural Networks (GNNs) are an effective framework for representation learning of graphs. GNNs follow a neighborhood aggregation scheme, where the representation vector of a node is computed by recursively aggregating and transforming representation vectors of its neighboring nodes. Many GNN variants have been proposed and have achieved state-of-the-art results on both node and graph classification tasks. However, despite GNNs revolutionizing graph representation learning, there is limited understanding of their representational properties and limitations. Here, we present a theoretical f","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks How Powerful are Graph Neural Networks? because it crossed a citation-hub threshold.","role_counts":[]},"graph":{"co_cited":[{"title":"Semi-Supervised Classification with Graph Convolutional Networks","work_id":"21fff118-807d-49cd-8229-f7087ba57b5d","shared_citers":18},{"title":"Graph Attention Networks","work_id":"7dd5bb04-b448-4f32-9719-5fc799641fd9","shared_citers":13},{"title":"ArXiv abs/2007.08663(2020)","work_id":"04426675-b2e3-4517-be09-08f2df683d14","shared_citers":4},{"title":"Inductive Representation Learning on Large Graphs","work_id":"a84007af-8130-4a77-b631-738c9af563fd","shared_citers":4},{"title":"Relational inductive biases, deep learning, and graph networks","work_id":"858410c0-7a66-4b27-b4e5-49aee9725be0","shared_citers":4},{"title":"Sdplib 1.2, a library of semidefinite program- ming test problems.Optimization Methods and Software, 11(1-4):683–690","work_id":"88e9cbb3-5e47-44fc-b1f0-f8bf82b9e07c","shared_citers":4},{"title":"arXiv preprint arXiv:1905.12265 , year=","work_id":"b6eca171-87f5-4e3b-8726-009d4d1b39d3","shared_citers":3},{"title":"Benchmarking graph neural networks","work_id":"87ffefbd-0e09-41bb-a247-fbc87d8d365e","shared_citers":3},{"title":"Neural message passing for quantum chemistry","work_id":"fb929706-486c-41f6-a1fd-96da310b2d62","shared_citers":3},{"title":"2010 , journal =","work_id":"1848424c-13f8-4544-a688-705f9b2422c2","shared_citers":2},{"title":"3d steerable cnns: Learning rotationally equivariant features in volumetric data.arXiv:1807.02547","work_id":"fbd44a4a-19a0-44c5-8f9e-43f5fc1d3ada","shared_citers":2},{"title":"Adam: A Method for Stochastic Optimization","work_id":"1910796d-9b52-4683-bf5c-de9632c1028b","shared_citers":2},{"title":"arXiv preprint arXiv:2203.06125 , year=","work_id":"f9af5e86-4da4-4d03-98e9-375c9dfb5dfa","shared_citers":2},{"title":"A Simple Structural Analysis Method for DAEs","work_id":"d8ce5cfc-20e8-456f-a19a-0b8698da903f","shared_citers":2},{"title":"Chemberta: Large-scale self- supervised pretraining for molecular property prediction","work_id":"44325e30-a45f-4052-8a05-3e9ed3f4c3cf","shared_citers":2},{"title":"Decoupled Weight Decay Regularization","work_id":"07ef7360-d385-4033-83f7-8384a6325204","shared_citers":2},{"title":"Fast Graph Representation Learning with PyTorch Geometric","work_id":"7030f6d8-1c95-4c18-a988-bfcf31d4cb4c","shared_citers":2},{"title":"Feinberg, Evan and Gomes, Joseph and Geniesse, Caleb and S","work_id":"6351d72d-6135-4484-afc8-abd85cfaa043","shared_citers":2},{"title":"How attentive are graph attention networks?arXiv preprint arXiv:2105.14491","work_id":"b81a77dc-03fc-40f2-a635-ba6c4a250268","shared_citers":2},{"title":"Invariant and equivariant graph networks.arXiv:1812.09902","work_id":"fb2e57cf-2565-45ea-8e6b-86753fbd52bf","shared_citers":2},{"title":"Kipf et al.Variational Graph Auto- Encoders","work_id":"2e8716eb-69bd-4617-8256-8543ba2c215c","shared_citers":2},{"title":"S., & Welling, M., 2016,Group Equivariant Convolu- tional Networks, Proc","work_id":"473fade8-b16b-45a0-b764-68a0f3957588","shared_citers":2},{"title":"Usbd: Universal structural basis distillation for source-free graph domain adaptation","work_id":"edabed69-11fa-4519-abcc-f26b69902a28","shared_citers":2},{"title":"Weidele, Claudio Bellei, Tom Robinson, and Charles E","work_id":"19e55ccd-457d-400e-b663-c70469ccb117","shared_citers":2}],"time_series":[{"n":34,"year":2026}],"dependency_candidates":[]},"authors":[]}}