Proposed by Google as an audience targeting scenario for Chrome, the “Federated Learning of Cohorts” (FLoC) creates lookalike audiences based on machine learning.
In the FLoC setup, cohort membership of a user is determined by the browser. Individual data is held by the browser and ad auction decisions are made in the browser.
According to GitHub, “The browser uses machine learning algorithms to develop a cohort based on the sites that an individual visits. The algorithms might be based on the URLs of the visited sites, on the content of those pages, or other factors.”