1: \begin{abstract}

2: In non-smooth stochastic optimization, we establish the non-convergence of the

3: stochastic subgradient descent (SGD) to the critical points recently called

4: active strict saddles by Davis and Drusvyatskiy.  Such points lie on a manifold

5: $M$ where the function $f$ has a direction of second-order negative curvature.

6: Off this manifold, the norm of the Clarke subdifferential of $f$ is

7: lower-bounded.  We require two conditions on $f$. The first assumption is a

8: Verdier stratification condition, which is a refinement of the popular Whitney

9: stratification.  It allows us to establish a strengthened version of the

10: projection formula of Bolte \emph{et al.} for Whitney stratifiable functions,

11: and which is of independent interest.  The second assumption, termed the angle

12: condition, allows to control the distance of the iterates to $M$.  When $f$ is

13: weakly convex, our assumptions are generic.  Consequently, generically in the

14: class of definable weakly convex functions, SGD converges to a local

15: minimizer.

16: \end{abstract}

17: