Patent · US Active

Training action-selection neural networks from demonstrations using multiple losses

US11604941B1 · kind B1 · utility

5Cited by

6References

20Claims

0Family size

Assignee

DeepMind Technologies Limited · GB

Inventor

Todd Hester · London, GB

Key dates

Filing date	Oct 29, 2018
Grant date	Mar 14, 2023
Priority date	—
Expiry date	Oct 13, 2041

Classification

Technology area (CPC G)Physics
CPC primaryG06V10/82
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

A method of training an action selection neural network to perform a demonstrated task using a supervised learning technique. The action selection neural network is configured to receive demonstration data comprising actions to perform the task and rewards received for performing the actions. The action selection neural network has auxiliary prediction task neural networks on one or more of its intermediate outputs. The action selection policy neural network is trained using multiple combined losses, concurrently with the auxiliary prediction task neural networks.

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