Deep relaxation of controlled stochastic gradient descent via singular perturbations

DOI10.1137/23M1544878zbMATH Open1542.93276MaRDI QIDQ6585233

Hicham Kouhkouh, Martino Bardi

Publication date: 9 August 2024

Published in: SIAM Journal on Control and Optimization (Search for Journal in Brave)

zbMATH Keywords

stochastic optimal controlhomogenizationHamilton-Jacobi-Bellman equationsmultiscale systemsdeep neural networksentropic stochastic gradient descent

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Time-scale analysis and singular perturbations in control/observation systems (93C70) Optimal stochastic control (93E20)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Optimal control with random parameters: a multiscale approach
• Invariant measures of diffusions with gradient drifts
• Weak convergence methods and singularly perturbed stochastic control and filtering problems
• On Poisson equation and diffusion approximation. II.
• On the Poisson equation and diffusion approximation. III
• On the Poisson equation and diffusion approximation. I
• McKean-Vlasov optimal control: the dynamic programming principle
• Deep relaxation: partial differential equations for optimizing deep neural networks
• The Kantorovich and variation distances between invariant measures of diffusions and nonlinear stationary Fokker-Planck-Kolmogorov equations
• Averaging of singularly perturbed controlled stochastic differential equations
• Controlled Markov processes and viscosity solutions
• Multiscale Stochastic Volatility for Equity, Interest Rate, and Credit Derivatives
• Uniqueness Results for Second-Order Bellman--Isaacs Equations under Quadratic Growth Assumptions and Applications
• Singular Perturbations in Ergodic Control of Diffusions
• Convergence by Viscosity Methods in Multiscale Financial Models with Stochastic Volatility
• A Generalization of Khasminskii's Theorem on the Existence of Invariant Measures for Locally Integrable Drifts
• Uniqueness of solutions of elliptic equations and uniqueness of invariant measures of diffusions
• Entropic gradient descent algorithms and wide flat minima*
• Entropy-SGD: biasing gradient descent into wide valleys
• Set-valued analysis
• Singular perturbations in stochastic optimal control with unbounded data

This page was built for publication: Deep relaxation of controlled stochastic gradient descent via singular perturbations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6585233)