{"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/boosting-with-lexicographic-programming","title":"Boosting with Lexicographic Programming: Addressing Class Imbalance without Cost Tuning","arxiv_id":"1708.09684","date":"2017-08-31","proceeding":null,"authors":["Shounak Datta","Sayak Nag","Swagatam Das"],"abstract":"A large amount of research effort has been dedicated to adapting boosting for\nimbalanced classification. However, boosting methods are yet to be\nsatisfactorily immune to class imbalance, especially for multi-class problems.\nThis is because most of the existing solutions for handling class imbalance\nrely on expensive cost set tuning for determining the proper level of\ncompensation. We show that the assignment of weights to the component\nclassifiers of a boosted ensemble can be thought of as a game of Tug of War\nbetween the classes in the margin space. We then demonstrate how this insight\ncan be used to attain a good compromise between the rare and abundant classes\nwithout having to resort to cost set tuning, which has long been the norm for\nimbalanced classification. The solution is based on a lexicographic linear\nprogramming framework which requires two stages. Initially, class-specific\ncomponent weight combinations are found so as to minimize a hinge loss\nindividually for each of the classes. Subsequently, the final component weights\nare assigned so that the maximum deviation from the class-specific minimum loss\nvalues (obtained in the previous stage) is minimized. Hence, the proposal is\nnot only restricted to two-class situations, but is also readily applicable to\nmulti-class problems. Additionally,we also derive the dual formulation\ncorresponding to the proposed framework. Experiments conducted on artificial\nand real-world imbalanced datasets as well as on challenging applications such\nas hyperspectral image classification and ImageNet classification establish the\nefficacy of the proposal.","url_abs":"http://arxiv.org/abs/1708.09684v2","url_pdf":"http://arxiv.org/pdf/1708.09684v2.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":"boosting-with-lexicographic-programming","repo_url":"https://github.com/Shounak-D/LexiBoost","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":"hyperspectral-image-classification","task_name":"Hyperspectral Image Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"},{"task_slug":"imbalanced-classification","task_name":"imbalanced 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}