There are use cases where the only accessible feedback for training machine-learning models is partial and biased (e.g., when feedback is obtained through surveys). When training machine-learning models on such data, there is a need to apply unbiasing techniques both during evaluation and learning. Damien Lefortier shares methods to handle these cases and explains how to ensure that they are performing well. Damien also offers an overview of the tools Facebook uses to apply these techniques.
Damien Lefortier is a machine-learning engineer on the Ads Ranking team at Facebook. Previously, Damien worked on the core Machine Learning team at Criteo, where he helped improve Criteo’s predictive algorithms for ad targeting, and on the Search team at Yandex, where he focused on search quality and infrastructure. He is working toward a PhD in information retrieval at the University of Amsterdam. His research work has been published at top tier conferences, such as WWW and CIKM.
Comments on this page are now closed.
©2017, O’Reilly UK Ltd • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • email@example.com