{"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/hopf-higher-order-propagation-framework-for","title":"HOPF: Higher Order Propagation Framework for Deep Collective Classification","arxiv_id":"1805.12421","date":"2018-05-31","proceeding":null,"authors":["Priyesh Vijayan","Yash Chandak","Mitesh M. Khapra","Srinivasan Parthasarathy","Balaraman Ravindran"],"abstract":"Given a graph where every node has certain attributes associated with it and\nsome nodes have labels associated with them, Collective Classification (CC) is\nthe task of assigning labels to every unlabeled node using information from the\nnode as well as its neighbors. It is often the case that a node is not only\ninfluenced by its immediate neighbors but also by higher order neighbors,\nmultiple hops away. Recent state-of-the-art models for CC learn end-to-end\ndifferentiable variations of Weisfeiler-Lehman (WL) kernels to aggregate\nmulti-hop neighborhood information. In this work, we propose a Higher Order\nPropagation Framework, HOPF, which provides an iterative inference mechanism\nfor these powerful differentiable kernels. Such a combination of classical\niterative inference mechanism with recent differentiable kernels allows the\nframework to learn graph convolutional filters that simultaneously exploit the\nattribute and label information available in the neighborhood. Further, these\niterative differentiable kernels can scale to larger hops beyond the memory\nlimitations of existing differentiable kernels. We also show that existing WL\nkernel-based models suffer from the problem of Node Information Morphing where\nthe information of the node is morphed or overwhelmed by the information of its\nneighbors when considering multiple hops. To address this, we propose a\nspecific instantiation of HOPF, called the NIP models, which preserves the node\ninformation at every propagation step. The iterative formulation of NIP models\nfurther helps in incorporating distant hop information concisely as summaries\nof the inferred labels. We do an extensive evaluation across 11 datasets from\ndifferent domains. We show that existing CC models do not provide consistent\nperformance across datasets, while the proposed NIP model with iterative\ninference is more robust.","url_abs":"http://arxiv.org/abs/1805.12421v6","url_pdf":"http://arxiv.org/pdf/1805.12421v6.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":"hopf-higher-order-propagation-framework-for","repo_url":"https://github.com/PriyeshV/HOPF","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"attribute","task_name":"Attribute"},{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}