This was part of Short Courses on the Mean Field Approach in Machine Learning and Statistics
Mean Field Asymptotics for High Dimensional Linear Models (Part 1)
Song Mei, University of California, Berkeley
Monday, October 18, 2021
Abstract: This course introduces the mean-field asymptotic results for high dimensional linear models, and discusses two applications in statistics and machine learning. One application is deriving the double-descent curve for the generalization error of over-parameterized models. The other application is analyzing the power of false discovery rate (FDR) controlling procedures.