1: \begin{abstract}

2: Stochastic Dual Coordinate Ascent is a popular method for

3: solving regularized loss minimization for the case of convex losses.

4: In this paper we show how a variant of SDCA can be applied for non-convex losses.

5: We prove linear convergence rate even if individual loss functions are

6: non-convex as long as the expected loss is convex. 

7: %Finally, we demonstrate the applicability to deep learning. 

8: \end{abstract}

9: