US 10,380,500 B2Granted

Version control for asynchronous distributed machine learning

AssigneeMicrosoft Technology Licensing, LLC

InventorsXu Miao, Yitong Zhou, Joel D. Young, Lijun Tang, Anmol Bhasin

Life of this patent

Estimating the term…

Loading patent…

Abstract

A system and method for managing asynchronously receiving updates and merging updates into global versions of a statistical model using version control are disclosed. During operation, the system transmits a first global version of a statistical model to a set of client computer systems. Next, the system obtains, from a first subset of the client computer systems, a first set of updates to the first global version. The system then merges the first set of updates into a second global version of the statistical model. Finally, the system transmits the second global version to the client computer systems asynchronously from receiving a second set of updates to the first and/or second global versions from a second subset of the client computer systems.

Claims17

Dependent claims

A method, comprising:

matching a server to a plurality of client computer systems for distribution of a statistical model based on a first attribute of the server and a second attribute of the plurality of client computer systems, wherein the first attribute comprises at least one of:
a processor speed;
an amount of memory;
and a network bandwidth;
transmitting, from the server to the plurality of client computer systems, a first global version of the statistical model;
obtaining, from a first subset of the plurality of client computer systems, a first set of updates to the first global version;
merging, by the server, the first set of updates into a second global version of the statistical model;
and
transmitting the second global version to one or more of the plurality of client computer systems asynchronously from receiving a second set of updates to one or more of the first and second global versions from a second subset of the plurality of client computer systems.

6 more elements

An apparatus, comprising:

one or more processors;
and
memory storing instructions that, when executed by the one or more processors, cause the apparatus to:
matching a server to a plurality of client computer systems for distribution of a statistical model based on a first attribute of the server and a second attribute of the plurality of client computer systems, wherein the first attribute comprises at least one of:
a processor speed;
an amount of memory;
and a network bandwidth;
transmit a first global version of the statistical model from the server to the plurality of client computer systems;
obtain, from a first subset of the plurality of client computer systems, a first set of updates to the first global version of the statistical model;
merge the first set of updates into a second global version of the statistical model;
and
transmit the second global version to one or more of the plurality of client computer systems asynchronously from receiving a second set of updates to one or more of the first and second global versions from a second subset of the plurality of client computer systems.

9 more elements

Dependencies are inferred from the claim text.