{"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/conic-multi-task-classification","title":"Conic Multi-Task Classification","arxiv_id":"1408.4714","date":"2014-08-20","proceeding":null,"authors":["Cong Li","Michael Georgiopoulos","Georgios C. Anagnostopoulos"],"abstract":"Traditionally, Multi-task Learning (MTL) models optimize the average of\ntask-related objective functions, which is an intuitive approach and which we\nwill be referring to as Average MTL. However, a more general framework,\nreferred to as Conic MTL, can be formulated by considering conic combinations\nof the objective functions instead; in this framework, Average MTL arises as a\nspecial case, when all combination coefficients equal 1. Although the advantage\nof Conic MTL over Average MTL has been shown experimentally in previous works,\nno theoretical justification has been provided to date. In this paper, we\nderive a generalization bound for the Conic MTL method, and demonstrate that\nthe tightest bound is not necessarily achieved, when all combination\ncoefficients equal 1; hence, Average MTL may not always be the optimal choice,\nand it is important to consider Conic MTL. As a byproduct of the generalization\nbound, it also theoretically explains the good experimental results of previous\nrelevant works. Finally, we propose a new Conic MTL model, whose conic\ncombination coefficients minimize the generalization bound, instead of choosing\nthem heuristically as has been done in previous methods. The rationale and\nadvantage of our model is demonstrated and verified via a series of experiments\nby comparing with several other methods.","url_abs":"http://arxiv.org/abs/1408.4714v1","url_pdf":"http://arxiv.org/pdf/1408.4714v1.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":"conic-multi-task-classification","repo_url":"https://github.com/congliucf/ECML2014","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"multi-task-learning","task_name":"Multi-Task 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}