Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Explore the first lecture in the TRIAD Distinguished Lecture Series featuring Sara van de Geer from ETH Zurich, focusing on Sharp Oracle Inequalities for Non-Convex Loss. Delve into the concept of M-estimation, a statistical procedure aimed at minimizing risk functions to achieve optimal data fit. Examine the role of regularization in preventing overfitting, with particular emphasis on the l₁-penalty and its application in the Lasso estimation procedure. Investigate the conditions necessary for penalties to yield sparsity oracle inequalities and their compatibility with risk functions. Gain insights into the theoretical foundations of statistical learning and optimization techniques in this 58-minute lecture presented by Georgia Tech Research.
