Patent · US Active

Distributional reinforcement learning for continuous control tasks

US11481629B2 · kind B2 · utility

1Cited by

0References

13Claims

0Family size

Assignee

DeepMind Technologies Limited · GB

Inventors

David Budden · London, GB
Matthew William Hoffman · London, GB
Gabriel Barth-Maron · London, GB

Key dates

Filing date	Oct 29, 2018
Grant date	Oct 25, 2022
Priority date	—
Expiry date	Jul 4, 2039

Classification

Technology area (CPC G)Physics
CPC primaryG06N3/044
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.