{"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/cost-sensitive-support-vector-machines","title":"Cost-Sensitive Support Vector Machines","arxiv_id":"1212.0975","date":"2012-12-05","proceeding":null,"authors":["Hamed Masnadi-Shirazi","Nuno Vasconcelos","Arya Iranmehr"],"abstract":"A new procedure for learning cost-sensitive SVM(CS-SVM) classifiers is\nproposed. The SVM hinge loss is extended to the cost sensitive setting, and the\nCS-SVM is derived as the minimizer of the associated risk. The extension of the\nhinge loss draws on recent connections between risk minimization and\nprobability elicitation. These connections are generalized to cost-sensitive\nclassification, in a manner that guarantees consistency with the cost-sensitive\nBayes risk, and associated Bayes decision rule. This ensures that optimal\ndecision rules, under the new hinge loss, implement the Bayes-optimal\ncost-sensitive classification boundary. Minimization of the new hinge loss is\nshown to be a generalization of the classic SVM optimization problem, and can\nbe solved by identical procedures. The dual problem of CS-SVM is carefully\nscrutinized by means of regularization theory and sensitivity analysis and the\nCS-SVM algorithm is substantiated. The proposed algorithm is also extended to\ncost-sensitive learning with example dependent costs. The minimum cost\nsensitive risk is proposed as the performance measure and is connected to ROC\nanalysis through vector optimization. The resulting algorithm avoids the\nshortcomings of previous approaches to cost-sensitive SVM design, and is shown\nto have superior experimental performance on a large number of cost sensitive\nand imbalanced datasets.","url_abs":"http://arxiv.org/abs/1212.0975v2","url_pdf":"http://arxiv.org/pdf/1212.0975v2.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":"cost-sensitive-support-vector-machines","repo_url":"https://github.com/BaptisteH3089/ImbLearning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"}],"methods":[{"method_slug":"svm","method_name":"SVM"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1212.0975","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}