Adaptive Stochastic Gradient Langevin Dynamics: Taming Convergence and Saddle Point Escape Time

Abstract
In this paper, we propose a new adaptive stochastic gradient Langevin dynamics (ASGLD) algorithmic framework and its two specialized versions, namely adaptive stochastic gradient (ASG) and adaptive gradient Langevin dynamics(AGLD), for non-convex optimization problems. All proposed algorithms can escape from saddle points with at most iterations, which is nearly dimension-free. Further, we show that ASGLD and ASG converge to a local minimum with at most iterations. Also, ASGLD with full gradients or ASGLD with a slowly linearly increasing batch size converge to a local minimum with iterations bounded by , which outperforms existing first-order methods.
View on arXivComments on this paper