{"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/orthogonal-machine-learning-power-and","title":"Orthogonal Machine Learning: Power and Limitations","arxiv_id":"1711.00342","date":"2017-11-01","proceeding":"ICML 2018 7","authors":["Lester Mackey","Vasilis Syrgkanis","Ilias Zadik"],"abstract":"Double machine learning provides $\\sqrt{n}$-consistent estimates of\nparameters of interest even when high-dimensional or nonparametric nuisance\nparameters are estimated at an $n^{-1/4}$ rate. The key is to employ\nNeyman-orthogonal moment equations which are first-order insensitive to\nperturbations in the nuisance parameters. We show that the $n^{-1/4}$\nrequirement can be improved to $n^{-1/(2k+2)}$ by employing a $k$-th order\nnotion of orthogonality that grants robustness to more complex or\nhigher-dimensional nuisance parameters. In the partially linear regression\nsetting popular in causal inference, we show that we can construct second-order\northogonal moments if and only if the treatment residual is not normally\ndistributed. Our proof relies on Stein's lemma and may be of independent\ninterest. We conclude by demonstrating the robustness benefits of an explicit\ndoubly-orthogonal estimation procedure for treatment effect.","url_abs":"http://arxiv.org/abs/1711.00342v6","url_pdf":"http://arxiv.org/pdf/1711.00342v6.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":"orthogonal-machine-learning-power-and","repo_url":"https://github.com/IliasZadik/double_orthogonal_ml","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"2k","task_name":"2k"},{"task_slug":"machine-learning","task_name":"BIG-bench Machine Learning"},{"task_slug":"causal-inference","task_name":"Causal Inference"},{"task_slug":"lemma","task_name":"LEMMA"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.00342","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}