Ranking and combining multiple predictors without labeled data

• Published in: Proceedings of the National Academy of Sciences - PNAS Vol. 111; no. 4; pp. 1253 - 1258
• Main Authors: Parisi, Fabio, Strino, Francesco, Nadler, Boaz, Kluger, Yuval
• Format: Journal Article
• Language: English
• Published: United States National Academy of Sciences 28.01.2014; National Acad Sciences
• Subjects: Approximation; Cartels; covariance; Covariance matrices; Datasets; Decision making; Eigenvectors; Estimate reliability; Initial guess; Likelihood Functions; Machine learning; Majority voting; Matrix; Maximum likelihood estimation; Maximum likelihood method; Models, Theoretical; Physical Sciences; prediction; Simulation; Test data; spectral analysis; classifier balanced accuracy; crowdsourcing; cartels; unsupervised learning

In a broad range of classification and decision-making problems, one is given the advice or predictions of several classifiers, of unknown reliability, over multiple questions or queries. This scenario is different from the standard supervised setting, where each classifier’s accuracy can be assesse...