Reinforcement Learning with Partial Parametric Model Knowledge

04/26/2023

∙

We adapt reinforcement learning (RL) methods for continuous control to bridge the gap between complete ignorance and perfect knowledge of the environment. Our method, Partial Knowledge Least Squares Policy Iteration (PLSPI), takes inspiration from both model-free RL and model-based control. It uses incomplete information from a partial model and retains RL's data-driven adaption towards optimal performance. The linear quadratic regulator provides a case study; numerical experiments demonstrate the effectiveness and resulting benefits of the proposed method.

READ FULL TEXT

Reinforcement Learning with Partial Parametric Model Knowledge

The Gap Between Model-Based and Model-Free Methods on the Linear Quadratic Regulator: An Asymptotic Viewpoint

Model-Based Policy Gradients with Parameter-Based Exploration by Least-Squares Conditional Density Estimation

Data efficient reinforcement learning and adaptive optimal perimeter control of network traffic dynamics

Model-based adaptation for sample efficient transfer in reinforcement learning control of parameter-varying systems

RL-Controller: a reinforcement learning framework for active structural control

A Modular and Transferable Reinforcement Learning Framework for the Fleet Rebalancing Problem

Policy Representation via Diffusion Probability Model for Reinforcement Learning

Reinforcement Learning with Partial Parametric Model Knowledge

Related Research

The Gap Between Model-Based and Model-Free Methods on the Linear Quadratic Regulator: An Asymptotic Viewpoint

Model-Based Policy Gradients with Parameter-Based Exploration by Least-Squares Conditional Density Estimation

Data efficient reinforcement learning and adaptive optimal perimeter control of network traffic dynamics

Model-based adaptation for sample efficient transfer in reinforcement learning control of parameter-varying systems

RL-Controller: a reinforcement learning framework for active structural control

A Modular and Transferable Reinforcement Learning Framework for the Fleet Rebalancing Problem

Policy Representation via Diffusion Probability Model for Reinforcement Learning