Wouter Koolen | Harder, Better, Faster, Stronger Convergence Rates for Least-Squares Regression ---- Dieuleveut, Flammarion, Bach

This will be the first Spring 2017 meeting of the CWI ML reading group. This paper is about first order algorithms (no inverse of no matrix) for stochastic least-squares regression that achieve the correct rate in terms of noise (d/n) and initial conditions (n^-2).