The reusable holdout: Preserving validity in adaptive data analysis

• DOI: 10.1126/science.aaa9375
• Publication Date: 2015-08-06T19:00:13Z
• Authors: Cynthia Dwork; Vitaly Feldman; Moritz Hardt; Toniann Pitassi; Omer Reingold; Aaron Roth
• Keywords: Spurious relationship; Data set; Data validation; Data Analysis

Testing hypotheses privately Large data sets offer a vast scope for testing already-formulated ideas and exploring new ones. Unfortunately, researchers who attempt to do both on the same set run risk of making false discoveries, even when exploration are carried out distinct subsets data. Based drawn from differential privacy, Dwork et al. now provide theoretical solution. Ideas tested against aggregate information, whereas individual components remain confidential. Preserving that privacy also preserves statistical inference validity. Science , this issue p. 636