{"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/contrastive-principal-component-analysis","title":"Contrastive Principal Component Analysis","arxiv_id":"1709.06716","date":"2017-09-20","proceeding":null,"authors":["Abubakar Abid","Martin J. Zhang","Vivek K. Bagaria","James Zou"],"abstract":"We present a new technique called contrastive principal component analysis\n(cPCA) that is designed to discover low-dimensional structure that is unique to\na dataset, or enriched in one dataset relative to other data. The technique is\na generalization of standard PCA, for the setting where multiple datasets are\navailable -- e.g. a treatment and a control group, or a mixed versus a\nhomogeneous population -- and the goal is to explore patterns that are specific\nto one of the datasets. We conduct a wide variety of experiments in which cPCA\nidentifies important dataset-specific patterns that are missed by PCA,\ndemonstrating that it is useful for many applications: subgroup discovery,\nvisualizing trends, feature selection, denoising, and data-dependent\nstandardization. We provide geometrical interpretations of cPCA and show that\nit satisfies desirable theoretical guarantees. We also extend cPCA to nonlinear\nsettings in the form of kernel cPCA. We have released our code as a python\npackage and documentation is on Github.","url_abs":"http://arxiv.org/abs/1709.06716v2","url_pdf":"http://arxiv.org/pdf/1709.06716v2.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":"contrastive-principal-component-analysis","repo_url":"https://github.com/abidlabs/contrastive","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"denoising","task_name":"Denoising"},{"task_slug":"subgroup-discovery","task_name":"Subgroup Discovery"},{"task_slug":"feature-selection","task_name":"feature selection"}],"methods":[{"method_slug":"pca","method_name":"PCA"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1709.06716","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}