Simultaneous Model Selection and Optimization through Parameter-free Stochastic Learning

Francesco Orabona

Stochastic gradient descent algorithms for training linear and kernel predictors are gaining more and more importance, thanks to their scalability. While various methods have been…