{"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/model-powered-conditional-independence-test","title":"Model-Powered Conditional Independence Test","arxiv_id":"1709.06138","date":"2017-09-18","proceeding":"NeurIPS 2017 12","authors":["Rajat Sen","Ananda Theertha Suresh","Karthikeyan Shanmugam","Alexandros G. Dimakis","Sanjay Shakkottai"],"abstract":"We consider the problem of non-parametric Conditional Independence testing\n(CI testing) for continuous random variables. Given i.i.d samples from the\njoint distribution $f(x,y,z)$ of continuous random vectors $X,Y$ and $Z,$ we\ndetermine whether $X \\perp Y | Z$. We approach this by converting the\nconditional independence test into a classification problem. This allows us to\nharness very powerful classifiers like gradient-boosted trees and deep neural\nnetworks. These models can handle complex probability distributions and allow\nus to perform significantly better compared to the prior state of the art, for\nhigh-dimensional CI testing. The main technical challenge in the classification\nproblem is the need for samples from the conditional product distribution\n$f^{CI}(x,y,z) = f(x|z)f(y|z)f(z)$ -- the joint distribution if and only if $X\n\\perp Y | Z.$ -- when given access only to i.i.d. samples from the true joint\ndistribution $f(x,y,z)$. To tackle this problem we propose a novel nearest\nneighbor bootstrap procedure and theoretically show that our generated samples\nare indeed close to $f^{CI}$ in terms of total variational distance. We then\ndevelop theoretical results regarding the generalization bounds for\nclassification for our problem, which translate into error bounds for CI\ntesting. We provide a novel analysis of Rademacher type classification bounds\nin the presence of non-i.i.d near-independent samples. We empirically validate\nthe performance of our algorithm on simulated and real datasets and show\nperformance gains over previous methods.","url_abs":"http://arxiv.org/abs/1709.06138v1","url_pdf":"http://arxiv.org/pdf/1709.06138v1.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":"model-powered-conditional-independence-test","repo_url":"https://github.com/rajatsen91/CCIT","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"generalization-bounds","task_name":"Generalization Bounds"},{"task_slug":"model","task_name":"model"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1709.06138","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}