{"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/multi-level-network-embedding-with-boosted","title":"Multi-Level Network Embedding with Boosted Low-Rank Matrix Approximation","arxiv_id":"1808.08627","date":"2018-08-26","proceeding":"ASONAM 2019 2019 11","authors":["Jundong Li","Liang Wu","Huan Liu"],"abstract":"As opposed to manual feature engineering which is tedious and difficult to\nscale, network representation learning has attracted a surge of research\ninterests as it automates the process of feature learning on graphs. The\nlearned low-dimensional node vector representation is generalizable and eases\nthe knowledge discovery process on graphs by enabling various off-the-shelf\nmachine learning tools to be directly applied. Recent research has shown that\nthe past decade of network embedding approaches either explicitly factorize a\ncarefully designed matrix to obtain the low-dimensional node vector\nrepresentation or are closely related to implicit matrix factorization, with\nthe fundamental assumption that the factorized node connectivity matrix is\nlow-rank. Nonetheless, the global low-rank assumption does not necessarily hold\nespecially when the factorized matrix encodes complex node interactions, and\nthe resultant single low-rank embedding matrix is insufficient to capture all\nthe observed connectivity patterns. In this regard, we propose a novel\nmulti-level network embedding framework BoostNE, which can learn multiple\nnetwork embedding representations of different granularity from coarse to fine\nwithout imposing the prevalent global low-rank assumption. The proposed BoostNE\nmethod is also in line with the successful gradient boosting method in ensemble\nlearning as multiple weak embeddings lead to a stronger and more effective one.\nWe assess the effectiveness of the proposed BoostNE framework by comparing it\nwith existing state-of-the-art network embedding methods on various datasets,\nand the experimental results corroborate the superiority of the proposed\nBoostNE network embedding framework.","url_abs":"http://arxiv.org/abs/1808.08627v1","url_pdf":"http://arxiv.org/pdf/1808.08627v1.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":"multi-level-network-embedding-with-boosted","repo_url":"https://github.com/benedekrozemberczki/BoostedFactorization","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"multi-level-network-embedding-with-boosted","repo_url":"https://github.com/benedekrozemberczki/karateclub","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"ensemble-learning","task_name":"Ensemble Learning"},{"task_slug":"feature-engineering","task_name":"Feature Engineering"},{"task_slug":"network-embedding","task_name":"Network Embedding"},{"task_slug":"representation-learning","task_name":"Representation Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}