公开(公告)号：US20210124739A1

公开(公告)日：2021-04-29

申请号：US16990506

申请日：2020-08-11

Applicant: Microsoft Technology Licensing, LLC

Inventor： Konstantinos KARANASOS ,  Matteo INTERLANDI ,  Fotios PSALLIDAS ,  Rathijit SEN ,  Kwanghyun PARK ,  Ivan POPIVANOV ,  Subramaniam VENKATRAMAN KRISHNAN ,  Markus WEIMER ,  Yuan YU ,  Raghunath RAMAKRISHNAN ,  Carlo Aldo CURINO ,  Doris Suiyi XIN ,  Karla Jean SAUR

IPC: G06F16/2458 ,  G06N5/04 ,  G06N20/00 ,  G06F16/28

Abstract: The description relates to executing an inference query relative to a database management system, such as a relational database management system. In one example a trained machine learning model can be stored within the database management system. An inference query can be received that applies the trained machine learning model on data local to the database management system. Analysis can be performed on the inference query and the trained machine learning model to generate a unified intermediate representation of the inference query and the trained model. Cross optimization can be performed on the unified intermediate representation. Based upon the cross-optimization, a first portion of the unified intermediate representation to be executed by a database engine of the database management system can be determined, and, a second portion of the unified intermediate representation to be executed by a machine learning runtime can be determined.

公开(公告)号：US20250156430A1

公开(公告)日：2025-05-15

申请号：US19022565

申请日：2025-01-15

Applicant: Microsoft Technology Licensing, LLC

Inventor： Konstantinos KARANASOS ,  Matteo INTERLANDI ,  Fotios PSALLIDAS ,  Rathijit SEN ,  Kwanghyun PARK ,  Ivan POPIVANOV ,  Subramaniam VENKATRAMAN KRISHNAN ,  Markus WEIMER ,  Yuan YU ,  Raghunath RAMAKRISHNAN ,  Carlo Aldo CURINO ,  Doris Suiyi XIN ,  Karla Jean SAUR

IPC: G06F16/2458 ,  G06F16/28 ,  G06N5/04 ,  G06N20/00

Abstract: The description relates to executing an inference query relative to a database management system, such as a relational database management system. In one example a trained machine learning model can be stored within the database management system. An inference query can be received that applies the trained machine learning model on data local to the database management system. Analysis can be performed on the inference query and the trained machine learning model to generate a unified intermediate representation of the inference query and the trained model. Cross optimization can be performed on the unified intermediate representation. Based upon the cross-optimization, a first portion of the unified intermediate representation to be executed by a database engine of the database management system can be determined, and, a second portion of the unified intermediate representation to be executed by a machine learning runtime can be determined.

公开(公告)号：US20230419166A1

公开(公告)日：2023-12-28

申请号：US17848679

申请日：2022-06-24

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： Devangkumar Rameshbhai PATEL ,  Wei ZUO ,  Yuan YU

IPC: G06N20/00 ,  G06K9/62 ,  G06T1/60

CPC classification number: G06N20/00 ,  G06K9/6256 ,  G06T1/60

Abstract: Some disclosed embodiments are directed to computing systems having different accelerators such that a first set of accelerators has a greater memory capability than a second set accelerators, while the second set of accelerators has a greater processing capability than the first set of accelerators. A machine learning model having different dense layers and sparse layers is distributed on the different accelerators such that the dense layers are distributed on one or more accelerators selected from the first set of accelerators and the sparse layers are distributed on one or more accelerators in the second set of accelerators.

公开(公告)号：US20230229905A1

公开(公告)日：2023-07-20

申请号：US17578326

申请日：2022-01-18

Applicant: Microsoft Technology Licensing, LLC

Inventor： Yuan YU

IPC: G06N3/08

CPC classification number: G06N3/08

Abstract: A method for training a machine-learning model. A plurality of nodes are assigned for training the machine-learning model. Nodes include agents comprising at least an agent processing unit and local memory. Each agent manages, via a local network, one or more workers that include a worker processing unit. Shards of a training data set are distributed for parallel processing by workers at different nodes. Each worker processing unit is configured to iteratively train on minibatches of a shard, and to report checkpoint states indicating updated parameters for storage in local memory. Based at least on recognizing a worker processing unit failing, the failed worker processing unit is reassigned and initialized based at least on a checkpoint state stored in local memory.