A projected primal-dual gradient optimal control method for deep reinforcement learning

DOI10.1186/S13362-020-00075-3zbMATH Open1472.49042OpenAlexW3029445142MaRDI QIDQ1980960FDOQ1980960

Authors: Simon Gottschalk, Michael Burger, Matthias Gerdts 

Publication date: 9 September 2021

Published in: Journal of Mathematics in Industry (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1186/s13362-020-00075-3

Recommendations

• Deep learning as optimal control problems: models and numerical methods
• Dynamical Systems andOptimal Control Approach to Deep Learning
• A mean-field optimal control formulation of deep learning
• Deep neural networks algorithms for stochastic control problems on finite horizon: convergence analysis
• A generalized path integral control approach to reinforcement learning

zbMATH Keywords

neural networksoptimal controlnecessary optimality conditionsreinforcement learningMarkov Decision Process

Mathematics Subject Classification ID

Applications of Markov chains and discrete-time Markov processes on general state spaces (social mobility, learning theory, industrial processes, etc.) (60J20) Optimality conditions for problems involving ordinary differential equations (49K15) Markov and semi-Markov decision processes (90C40) Stochastic learning and adaptive control (93E35) Networks and circuits as models of computation; circuit complexity (68Q06)

Cites Work

• \({\mathcal Q}\)-learning
• Robust optimization
• Title not available (Why is that?)
• Optimal control of ODEs and DAEs.
• Title not available (Why is that?)
• Simple statistical gradient-following algorithms for connectionist reinforcement learning
• Reinforcement learning. An introduction
• Title not available (Why is that?)
• Handbook of Markov decision processes. Methods and applications
• Title not available (Why is that?)
• Ordinary differential equations. An introduction from the dynamical systems perspective
• Deep learning as optimal control problems: models and numerical methods
• Reinforcement Learning Applied to a Human Arm Model

Cited In (5)

• A mean-field optimal control formulation of deep learning
• Value-Gradient Based Formulation of Optimal Control Problem and Machine Learning Algorithm
• Primal-Dual Q-Learning Framework for LQR Design
• Pretty darn good control: when are approximate solutions better than approximate models
• Jointly learning environments and control policies with projected stochastic gradient ascent

This page was built for publication: A projected primal-dual gradient optimal control method for deep reinforcement learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1980960)