Nonasymptotic estimates for stochastic gradient Langevin dynamics under local conditions in nonconvex optimization

DOI10.1007/s00245-022-09932-6OpenAlexW4316041201MaRDI QIDQ2682367

Sotirios Sabanis, Ying Zhang, Theodoros Damoulas, Ömer Deniz Akyildiz

Publication date: 31 January 2023

Published in: Applied Mathematics and Optimization (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1910.02008

zbMATH Keywords

non-convex optimizationvariational inferencestochastic gradient Langevin dynamicslocal Lipschitz continuouslocal dissipativitynon-asymptotic estimates

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Sampling theory, sample surveys (62D05) Monte Carlo methods (65C05) Nonconvex programming, global optimization (90C26) Applications of Markov chains and discrete-time Markov processes on general state spaces (social mobility, learning theory, industrial processes, etc.) (60J20) Numerical analysis or methods applied to Markov chains (65C40)

Related Items (5)

Statistical Finite Elements via Langevin Dynamics ⋮ Global Optimization via Schrödinger–Föllmer Diffusion ⋮ Distributed event-triggered unadjusted Langevin algorithm for Bayesian learning ⋮ Taming Neural Networks with TUSLA: Nonconvex Learning via Adaptive Stochastic Gradient Langevin Algorithms ⋮ Multi-index antithetic stochastic gradient algorithm

Cites Work

• Reflection couplings and contraction rates for diffusions
• Optimal portfolio selection and dynamic benchmark tracking
• Fixed-form variational posterior approximation through stochastic linear regression
• Laplace's method revisited: Weak convergence of probability measures
• On stochastic gradient Langevin dynamics with dependent data streams in the logconcave case
• The tamed unadjusted Langevin algorithm
• User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient
• High-dimensional Bayesian inference via the unadjusted Langevin algorithm
• Higher order Langevin Monte Carlo algorithm
• Nonasymptotic convergence analysis for the unadjusted Langevin algorithm
• Ergodicity for SDEs and approximations: locally Lipschitz vector fields and degenerate noise.
• Graphical Models, Exponential Families, and Variational Inference
• A useful theorem for nonlinear devices having Gaussian inputs
• Quantitative Harris-type theorems for diffusions and McKean–Vlasov processes
• On fixed gain recursive estimators with discontinuity in the parameters
• On Stochastic Gradient Langevin Dynamics with Dependent Data Streams: The Fully Nonconvex Case
• Theoretical Guarantees for Approximate Sampling from Smooth and Log-Concave Densities

This page was built for publication: Nonasymptotic estimates for stochastic gradient Langevin dynamics under local conditions in nonconvex optimization