Deep Learning Approximation for Stochastic Control Problems
read the original abstract
Many real world stochastic control problems suffer from the "curse of dimensionality". To overcome this difficulty, we develop a deep learning approach that directly solves high-dimensional stochastic control problems based on Monte-Carlo sampling. We approximate the time-dependent controls as feedforward neural networks and stack these networks together through model dynamics. The objective function for the control problem plays the role of the loss function for the deep neural network. We test this approach using examples from the areas of optimal trading and energy storage. Our results suggest that the algorithm presented here achieves satisfactory accuracy and at the same time, can handle rather high dimensional problems.
This paper has not been read by Pith yet.
Forward citations
Cited by 5 Pith papers
-
Dual Approaches to Stochastic Control via SPDEs and the Pathwise Hopf Formula
The authors prove the generalized Hopf formula under mild conditions and use it with SPDEs and the Pontryagin principle to compute curse-of-dimensionality-free dual bounds for stochastic control.
-
PI-SONet: A Physics-Informed Symplectic Operator Network for Real-Time Optimal Control of Multi-Agent Systems
PI-SONet trains a single structure-preserving operator network to deliver sub-second approximations to Pontryagin Maximum Principle solutions for parameterized multi-agent optimal control problems.
-
A Deep Ritz Method for High-Dimensional Steady States of the Cahn-Hilliard Equation
A Deep Ritz method with augmented Lagrangian and Fourier feature mappings computes high-dimensional steady states of the Cahn-Hilliard equation and identifies multiple nontrivial phase separation patterns.
-
Adversarial Decision-Making in Partially Observable Multi-Agent Systems: A Sequential Hypothesis Testing Approach
Presents an SHT-driven framework for adversarial decision-making in partially observable multi-agent systems formulated as a partially observable Stackelberg game, with semi-explicit optimal controls for the blue team...
-
Neural Actor-Critic Methods for Hamilton-Jacobi-Bellman PDEs: Asymptotic Analysis and Numerical Studies
Neural actor-critic method for high-dimensional HJB PDEs converges in Sobolev space to an infinite-dimensional ODE whose fixed points solve the stochastic control problem under a convexity-like Hamiltonian assumption,...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.