Non-convex learning via Stochastic Gradient Langevin Dynamics: a nonasymptotic analysis.

Maxim Raginsky, Alexander Rakhlin, Matus Telgarsky

Browse the full COLT paper archive.