The patent badge is an abbreviated version of the USPTO patent document. The patent badge does contain a link to the full patent document.

The patent badge is an abbreviated version of the USPTO patent document. The patent badge covers the following: Patent number, Date patent was issued, Date patent was filed, Title of the patent, Applicant, Inventor, Assignee, Attorney firm, Primary examiner, Assistant examiner, CPCs, and Abstract. The patent badge does contain a link to the full patent document (in Adobe Acrobat format, aka pdf). To download or print any patent click here.

Date of Patent:
Dec. 27, 2022

Filed:

Oct. 24, 2018
Applicant:

Magic Leap, Inc., Plantation, FL (US);

Inventors:

Zhao Chen, Sunnyvale, CA (US);

Vijay Badrinarayanan, Mountain View, CA (US);

Andrew Rabinovich, San Francisco, CA (US);

Assignee:

Magic Leap, Inc., Plantation, FL (US);

Attorney:
Primary Examiner:
Assistant Examiner:
Int. Cl.
CPC ...
G06N 3/08 (2006.01); G06N 3/04 (2006.01); G06N 20/00 (2019.01);
U.S. Cl.
CPC ...
G06N 3/084 (2013.01); G06N 3/0472 (2013.01); G06N 20/00 (2019.01);
Abstract

Systems and methods for training a multitask network is disclosed. In one aspect, training the multitask network includes determining a gradient norm of a single-task loss adjusted by a task weight for each task, with respect to network weights of the multitask network, and a relative training rate for the task based on the single-task loss for the task. Subsequently, a gradient loss function, comprising a difference between (1) the determined gradient norm for each task and (2) a corresponding target gradient norm, can be determined. An updated task weight for the task can be determined and used in the next iteration of training the multitask network, using a gradient of the gradient loss function with respect to the task weight for the task.


Find Patent Forward Citations

Loading…