Non-Convex Optimization via Non-Reversible Stochastic Gradient Langevin Dynamics

Stochastic Gradient Langevin Dynamics (SGLD) is a powerful algorithm for optimizing a non-convex objective, where a controlled and properly scaled Gaussian noise is added to the stochastic gradients to steer the iterates towards a global minimum. SGLD is based on the overdamped Langevin diffusion wh...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2020-06
Main Authors: Hu, Yuanhan, Wang, Xiaoyu, Gao, Xuefeng, Gurbuzbalaban, Mert, Zhu, Lingjiong
Format: Article
Language:English
Subjects:
Algorithms
Computational geometry
Convergence
Convexity
Diffusion rate
Empirical analysis
Independent component analysis
Mathematical analysis
Matrix methods
Neural networks
Optimization
Polynomials
Random noise
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Stochastic Gradient Langevin Dynamics (SGLD) is a powerful algorithm for optimizing a non-convex objective, where a controlled and properly scaled Gaussian noise is added to the stochastic gradients to steer the iterates towards a global minimum. SGLD is based on the overdamped Langevin diffusion which is reversible in time. By adding an anti-symmetric matrix to the drift term of the overdamped Langevin diffusion, one gets a non-reversible diffusion that converges to the same stationary distribution with a faster convergence rate. In this paper, we study the non reversible Stochastic Gradient Langevin Dynamics (NSGLD) which is based on discretization of the non-reversible Langevin diffusion. We provide finite-time performance bounds for the global convergence of NSGLD for solving stochastic non-convex optimization problems. Our results lead to non-asymptotic guarantees for both population and empirical risk minimization problems. Numerical experiments for Bayesian independent component analysis and neural network models show that NSGLD can outperform SGLD with proper choices of the anti-symmetric matrix.
ISSN:2331-8422