Session A7 - Stochastic Computation
July 12, 18:00 ~ 18:25 - Room B2
Stochastic Composite Least-Squares Regression with convergence rate $O(1/n)$
INRIA - Ecole Normale Supérieure, France - firstname.lastname@example.org
We consider the minimization of composite objective functions composed of the expectation of quadratic functions and an arbitrary convex function. We study the stochastic dual averaging algorithm with a constant step-size, showing that it leads to a convergence rate of $O(1/n)$ without strong convexity assumptions. This thus extends earlier results on least-squares regression with the Euclidean geometry to (a) all convex regularizers and constraints, and (b) all geometries represented by a Bregman divergence. This is achieved by a new proof technique that relates stochastic and deterministic recursions.
Joint work with Nicolas Flammarion.