
Learning without Concentration
Published on Feb 4, 20254985 Views
We obtain sharp bounds on the convergence rate of Empirical Risk Minimization performed in a convex class and with respect to the squared loss, without any boundedness assumptions on class members or