{"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/lanczosnet-multi-scale-deep-graph","title":"LanczosNet: Multi-Scale Deep Graph Convolutional Networks","arxiv_id":"1901.01484","date":"2019-01-06","proceeding":"ICLR 2019 5","authors":["Renjie Liao","Zhizhen Zhao","Raquel Urtasun","Richard S. Zemel"],"abstract":"We propose the Lanczos network (LanczosNet), which uses the Lanczos algorithm to construct low rank approximations of the graph Laplacian for graph convolution. Relying on the tridiagonal decomposition of the Lanczos algorithm, we not only efficiently exploit multi-scale information via fast approximated computation of matrix power but also design learnable spectral filters. Being fully differentiable, LanczosNet facilitates both graph kernel learning as well as learning node embeddings. We show the connection between our LanczosNet and graph based manifold learning methods, especially the diffusion maps. We benchmark our model against several recent deep graph networks on citation networks and QM8 quantum chemistry dataset. Experimental results show that our model achieves the state-of-the-art performance in most tasks. Code is released at: \\url{https://github.com/lrjconan/LanczosNetwork}.","url_abs":"https://arxiv.org/abs/1901.01484v2","url_pdf":"https://arxiv.org/pdf/1901.01484v2.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":"lanczosnet-multi-scale-deep-graph","repo_url":"https://github.com/lrjconan/LanczosNetwork","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"node-classification","task_name":"Node Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/node-classification-on-citeseer-05","task":"Node Classification","dataset":"CiteSeer (0.5%)","model":"AdaLanczosNet","rank_in_archive_order":7,"of":14,"metrics":{"Accuracy":"53.8 ± 4.7"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-citeseer-05","task":"Node Classification","dataset":"CiteSeer (0.5%)","model":"LanczosNet","rank_in_archive_order":8,"of":14,"metrics":{"Accuracy":"53.2 ± 4.0"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-citeseer-1","task":"Node Classification","dataset":"CiteSeer (1%)","model":"AdaLanczosNet","rank_in_archive_order":7,"of":14,"metrics":{"Accuracy":"63.3 ± 1.8"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-citeseer-1","task":"Node Classification","dataset":"CiteSeer (1%)","model":"LanczosNet","rank_in_archive_order":9,"of":14,"metrics":{"Accuracy":"61.3 ± 3.9"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-citeseer-with-public","task":"Node Classification","dataset":"CiteSeer with Public Split: fixed 20 nodes per class","model":"AdaLanczosNet","rank_in_archive_order":35,"of":40,"metrics":{"Accuracy":"68.7 ± 1.0"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-citeseer-with-public","task":"Node Classification","dataset":"CiteSeer with Public Split: fixed 20 nodes per class","model":"LanczosNet","rank_in_archive_order":37,"of":40,"metrics":{"Accuracy":"66.2 ± 1.9"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-05","task":"Node Classification","dataset":"Cora (0.5%)","model":"AdaLanczosNet","rank_in_archive_order":8,"of":15,"metrics":{"Accuracy":"60.8 ± 9.0"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-05","task":"Node Classification","dataset":"Cora (0.5%)","model":"LanczosNet","rank_in_archive_order":10,"of":15,"metrics":{"Accuracy":"58.1 ± 8.2"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-1","task":"Node Classification","dataset":"Cora (1%)","model":"AdaLanczosNet","rank_in_archive_order":8,"of":15,"metrics":{"Accuracy":"67.5 ± 8.7"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-1","task":"Node Classification","dataset":"Cora (1%)","model":"LanczosNet","rank_in_archive_order":10,"of":15,"metrics":{"Accuracy":"66.1 ± 8.2"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-3","task":"Node Classification","dataset":"Cora (3%)","model":"AdaLanczosNet","rank_in_archive_order":8,"of":15,"metrics":{"Accuracy":"77.7 ± 2.4"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-3","task":"Node Classification","dataset":"Cora (3%)","model":"LanczosNet","rank_in_archive_order":10,"of":15,"metrics":{"Accuracy":"76.3 ± 2.3"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-with-public-split","task":"Node Classification","dataset":"Cora with Public Split: fixed 20 nodes per class","model":"AdaLanczosNet","rank_in_archive_order":30,"of":36,"metrics":{"Accuracy":"80.4 ± 1.1"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-cora-with-public-split","task":"Node Classification","dataset":"Cora with Public Split: fixed 20 nodes per class","model":"LanczosNet","rank_in_archive_order":32,"of":36,"metrics":{"Accuracy":"79.5 ± 1.8"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-003","task":"Node Classification","dataset":"PubMed (0.03%)","model":"AdaLanczosNet","rank_in_archive_order":7,"of":14,"metrics":{"Accuracy":"61%"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-003","task":"Node Classification","dataset":"PubMed (0.03%)","model":"LanczosNet","rank_in_archive_order":9,"of":14,"metrics":{"Accuracy":"60.4 ± 8.6"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-005","task":"Node Classification","dataset":"PubMed (0.05%)","model":"LanczosNet","rank_in_archive_order":7,"of":14,"metrics":{"Accuracy":"68.8 ± 5.6"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-005","task":"Node Classification","dataset":"PubMed (0.05%)","model":"AdaLanczosNet","rank_in_archive_order":9,"of":14,"metrics":{"Accuracy":"66%"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-01","task":"Node Classification","dataset":"PubMed (0.1%)","model":"LanczosNet","rank_in_archive_order":6,"of":14,"metrics":{"Accuracy":"73.4 ± 5.1"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-01","task":"Node Classification","dataset":"PubMed (0.1%)","model":"AdaLanczosNet","rank_in_archive_order":9,"of":14,"metrics":{"Accuracy":"72.8 ± 4.6"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-with-public","task":"Node Classification","dataset":"PubMed with Public Split: fixed 20 nodes per class","model":"LanczosNet","rank_in_archive_order":27,"of":37,"metrics":{"Accuracy":"78.3 ± 0.3"},"uses_additional_data":false},{"leaderboard":"/sota/node-classification-on-pubmed-with-public","task":"Node Classification","dataset":"PubMed with Public Split: fixed 20 nodes per class","model":"AdaLanczosNet","rank_in_archive_order":28,"of":37,"metrics":{"Accuracy":"78.1 ± 0.4"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1901.01484","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}