{"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/canonical-correlation-forests","title":"Canonical Correlation Forests","arxiv_id":"1507.05444","date":"2015-07-20","proceeding":null,"authors":["Tom Rainforth","Frank Wood"],"abstract":"We introduce canonical correlation forests (CCFs), a new decision tree\nensemble method for classification and regression. Individual canonical\ncorrelation trees are binary decision trees with hyperplane splits based on\nlocal canonical correlation coefficients calculated during training. Unlike\naxis-aligned alternatives, the decision surfaces of CCFs are not restricted to\nthe coordinate system of the inputs features and therefore more naturally\nrepresent data with correlated inputs. CCFs naturally accommodate multiple\noutputs, provide a similar computational complexity to random forests, and\ninherit their impressive robustness to the choice of input parameters. As part\nof the CCF training algorithm, we also introduce projection bootstrapping, a\nnovel alternative to bagging for oblique decision tree ensembles which\nmaintains use of the full dataset in selecting split points, often leading to\nimprovements in predictive accuracy. Our experiments show that, even without\nparameter tuning, CCFs out-perform axis-aligned random forests and other\nstate-of-the-art tree ensemble methods on both classification and regression\nproblems, delivering both improved predictive accuracy and faster training\ntimes. We further show that they outperform all of the 179 classifiers\nconsidered in a recent extensive survey.","url_abs":"http://arxiv.org/abs/1507.05444v6","url_pdf":"http://arxiv.org/pdf/1507.05444v6.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":"canonical-correlation-forests","repo_url":"https://github.com/twgr/ccfs","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"canonical-correlation-forests","repo_url":"https://github.com/plai-group/ccfs-python","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"canonical-correlation-forests","repo_url":"https://github.com/tonyjo/ccfs-python","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}