{"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/distill-and-compare-auditing-black-box-models","title":"Distill-and-Compare: Auditing Black-Box Models Using Transparent Model Distillation","arxiv_id":"1710.06169","date":"2017-10-17","proceeding":null,"authors":["Sarah Tan","Rich Caruana","Giles Hooker","Yin Lou"],"abstract":"Black-box risk scoring models permeate our lives, yet are typically\nproprietary or opaque. We propose Distill-and-Compare, a model distillation and\ncomparison approach to audit such models. To gain insight into black-box\nmodels, we treat them as teachers, training transparent student models to mimic\nthe risk scores assigned by black-box models. We compare the student model\ntrained with distillation to a second un-distilled transparent model trained on\nground-truth outcomes, and use differences between the two models to gain\ninsight into the black-box model. Our approach can be applied in a realistic\nsetting, without probing the black-box model API. We demonstrate the approach\non four public data sets: COMPAS, Stop-and-Frisk, Chicago Police, and Lending\nClub. We also propose a statistical test to determine if a data set is missing\nkey features used to train the black-box model. Our test finds that the\nProPublica data is likely missing key feature(s) used in COMPAS.","url_abs":"http://arxiv.org/abs/1710.06169v4","url_pdf":"http://arxiv.org/pdf/1710.06169v4.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":"distill-and-compare-auditing-black-box-models","repo_url":"https://github.com/shftan/auditblackbox","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1710.06169","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}