Abstract:
Prediction-powered inference is a framework for performing valid statistical inference when a gold-standard data set is supplemented with predictions from a machine-learning system. The framework yields simple algorithms for computing provably valid confidence intervals for quantities such as means, quantiles, and linear and logistic regression coefficients, without making any assumptions about the machine-learning algorithm that provides the predictions. Furthermore, more accurate predictions yield smaller confidence intervals. Prediction-powered inference may enable scientists to draw valid and more data-efficient conclusions using machine learning.
Bio:
Clara Wong-Fannjiang is a machine learning scientist at Genentech, Prescient Frontier Research. She received her Ph.D. in Electrical Engineering and Computer Sciences from UC Berkeley in August 2023, advised by Michael I. Jordan and Jennifer Listgarten. She works on methods for trustworthy scientific inquiry using machine learning, particularly in the context of drug discovery. Prior to Berkeley, she was a Google AI resident and researcher at the Monterey Bay Aquarium Research Institute, and received her B.S. in Computer Science from Stanford University.
21 окт 2023
