Abstract: Availability of large high-dimensional data-sets has urged the  development of optimization solutions for large scale learning problems.  From a theoretical perspective this has motivated the goal of better  understanding the interplay between statistics and optimization, towards developing new, more efficient  learning algorithms. Indeed,  while much theoretical work has been  devoted to study  statistical properties of estimators defined by variational schemes (a.k.a. Tikhonov regularization),  and  the  computational  properties of optimization procedures to solve the corresponding minimization problems, much less  work has  been devoted to  the integration of statistical and optimization aspects.  In this talk, we will present some recent proposals to develop machine learning algorithms which are provably efficient as well as statistically sound.  In particular, we will discuss different instances of iterative regularization methods and, if time permit,  randomized sampling techniques allowing further improvements. Short BIO: http://web.mit.edu/lrosasco/www/