Policy gradient in Lipschitz Markov decision processes

DOI10.1007/s10994-015-5484-1zbMath1354.90166OpenAlexW2046859786MaRDI QIDQ747252

Matteo Pirotta, Luca Bascetta, Marcello Restelli

Publication date: 23 October 2015

Published in: Machine Learning (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1007/s10994-015-5484-1

zbMATH Keywords

Lipschitz continuityMarkov decision processreinforcement learningpolicy gradient algorithm

Mathematics Subject Classification ID

Markov and semi-Markov decision processes (90C40) Methods of reduced gradient type (90C52)

Related Items

Unnamed Item ⋮ A Small Gain Analysis of Single Timescale Actor Critic ⋮ Smoothing policies and safe policy gradients ⋮ Risk-averse optimization of reward-based coherent risk measures ⋮ On the sample complexity of actor-critic method for reinforcement learning with function approximation ⋮ Global Convergence of Policy Gradient Methods to (Almost) Locally Optimal Policies ⋮ Unnamed Item

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Policy search for motor primitives in robotics
• Lipschitz continuity of value functions in Markovian decision processes
• Stochastic optimal control. The discrete time case
• Minimization of functions having Lipschitz continuous first partial derivatives
• Collective motions of a shell structure
• Multivariate stochastic approximation using a simultaneous perturbation gradient approximation
• Line search algorithms with guaranteed sufficient decrease
• A Stochastic Approximation Method
• Solving connection and linearization problems within the Askey scheme and its \(q\)-analogue via inversion formulas

This page was built for publication: Policy gradient in Lipschitz Markov decision processes