Recent advances in statistical learning and convex optimization have inspired many successful practices. Standard theories assume smoothness---bounded gradient, Hessian, etc.---and strong convexity of the loss function. Unfortunately, such conditions may ...
EPFL2018