{"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/a-hybrid-algorithm-for-bayesian-network","title":"A hybrid algorithm for Bayesian network structure learning with application to multi-label learning","arxiv_id":"1506.05692","date":"2015-06-18","proceeding":null,"authors":["Maxime Gasse","Alex Aussem","Haytham Elghazel"],"abstract":"We present a novel hybrid algorithm for Bayesian network structure learning,\ncalled H2PC. It first reconstructs the skeleton of a Bayesian network and then\nperforms a Bayesian-scoring greedy hill-climbing search to orient the edges.\nThe algorithm is based on divide-and-conquer constraint-based subroutines to\nlearn the local structure around a target variable. We conduct two series of\nexperimental comparisons of H2PC against Max-Min Hill-Climbing (MMHC), which is\ncurrently the most powerful state-of-the-art algorithm for Bayesian network\nstructure learning. First, we use eight well-known Bayesian network benchmarks\nwith various data sizes to assess the quality of the learned structure returned\nby the algorithms. Our extensive experiments show that H2PC outperforms MMHC in\nterms of goodness of fit to new data and quality of the network structure with\nrespect to the true dependence structure of the data. Second, we investigate\nH2PC's ability to solve the multi-label learning problem. We provide\ntheoretical results to characterize and identify graphically the so-called\nminimal label powersets that appear as irreducible factors in the joint\ndistribution under the faithfulness condition. The multi-label learning problem\nis then decomposed into a series of multi-class classification problems, where\neach multi-class variable encodes a label powerset. H2PC is shown to compare\nfavorably to MMHC in terms of global classification accuracy over ten\nmulti-label data sets covering different application domains. Overall, our\nexperiments support the conclusions that local structural learning with H2PC in\nthe form of local neighborhood induction is a theoretically well-motivated and\nempirically effective learning framework that is well suited to multi-label\nlearning. The source code (in R) of H2PC as well as all data sets used for the\nempirical tests are publicly available.","url_abs":"http://arxiv.org/abs/1506.05692v1","url_pdf":"http://arxiv.org/pdf/1506.05692v1.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":"a-hybrid-algorithm-for-bayesian-network","repo_url":"https://github.com/madbix/bnlearn-clone-3.4","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"multi-label-learning","task_name":"Multi-Label Learning"},{"task_slug":"multi-class-classification","task_name":"Multi-class Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1506.05692","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}