Training action selection neural networks using apprenticeship

US11468321B2 · kind B2 · utility

Assignee

• DeepMind Technologies Limited · GB

Inventors

• Olivier Claude Pietquin · Lille, FR
• Martin Riedmiller · Balgheim, DE
• Wang Fumin · London, GB
• Bilal Piot · London, GB
• Mel Vecerik · London, GB
• Todd Hester · London, GB
• Thomas Rothoerl · London, GB
• Thomas Lampe · London, GB
• Nicolas Manfred Otto Heess · London, GB
• Jonathan Karl Scholz · London, GB

Key dates

• Filing date: Jun 28, 2018
• Grant date: Oct 11, 2022
• Expiry date: Jul 1, 2039

Classification

• Technology area (CPC —): General

Abstract

An off-policy reinforcement learning actor-critic neural network system configured to select actions from a continuous action space to be performed by an agent interacting with an environment to perform a task. An observation defines environment state data and reward data. The system has an actor neural network which learns a policy function mapping the state data to action data. A critic neural network learns an action-value (Q) function. A replay buffer stores tuples of the state data, the action data, the reward data and new state data. The replay buffer also includes demonstration transition data comprising a set of the tuples from a demonstration of the task within the environment. The neural network system is configured to train the actor neural network and the critic neural network off-policy using stored tuples from the replay buffer comprising tuples both from operation of the system and from the demonstration transition data.