公开(公告)号：US10289962B2

公开(公告)日：2019-05-14

申请号：US14731349

申请日：2015-06-04

Applicant: Google LLC

Inventor： Oriol Vinyals ,  Jeffrey A. Dean ,  Geoffrey E. Hinton

IPC: G06N3/04 ,  G06N7/00 ,  G06N99/00

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a distilled machine learning model. One of the methods includes training a cumbersome machine learning model, wherein the cumbersome machine learning model is configured to receive an input and generate a respective score for each of a plurality of classes; and training a distilled machine learning model on a plurality of training inputs, wherein the distilled machine learning model is also configured to receive inputs and generate scores for the plurality of classes, comprising: processing each training input using the cumbersome machine learning model to generate a cumbersome target soft output for the training input; and training the distilled machine learning model to, for each of the training inputs, generate a soft output that matches the cumbersome target soft output for the training input.

公开(公告)号：US10733535B1

公开(公告)日：2020-08-04

申请号：US15665236

申请日：2017-07-31

Applicant: Google LLC

Inventor： Gregory S. Corrado ,  Kai Chen ,  Jeffrey A. Dean ,  Samy Bengio ,  Rajat Monga ,  Matthieu Devin

IPC: G06F15/18 ,  G06N20/00 ,  G06N7/08 ,  G06K9/62 ,  G06N7/00 ,  G06N5/02

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a model using parameter server shards. One of the methods includes receiving, at a parameter server shard configured to maintain values of a disjoint partition of the parameters of the model, a succession of respective requests for parameter values from each of a plurality of replicas of the model; in response to each request, downloading a current value of each requested parameter to the replica from which the request was received; receiving a succession of uploads, each upload including respective delta values for each of the parameters in the partition maintained by the shard; and updating values of the parameters in the partition maintained by the parameter server shard repeatedly based on the uploads of delta values to generate current parameter values.

公开(公告)号：US20190243804A1

公开(公告)日：2019-08-08

申请号：US16268149

申请日：2019-02-05

Applicant: Google LLC

Inventor： Yasushi Saito ,  Sanjay Ghemawat ,  Jeffrey A. Dean

IPC: G06F16/16 ,  G06F16/174 ,  G06F16/11 ,  G06F16/182 ,  G06F16/215

CPC classification number: G06F16/162 ,  G06F16/11 ,  G06F16/1748 ,  G06F16/182 ,  G06F16/215

Abstract: A method for deleting obsolete files from a file system is provided. The method includes receiving a request to delete a reference to a first target file of a plurality of target files stored in a file system, the first target file having a first target file name. A first reference file whose file name includes the first target file name is identified. The first reference file is deleted from the file system. The method further includes determining whether the file system includes at least one reference file, distinct from the first reference file, whose file name includes the first target file name. In accordance with a determination that the file system does not include the at least one reference file, the first target file is deleted from the file system.

公开(公告)号：US10241997B1

公开(公告)日：2019-03-26

申请号：US15682374

申请日：2017-08-21

Applicant: Google LLC

Inventor： Tomas Mikolov ,  Kai Chen ,  Gregory S. Corrado ,  Jeffrey A. Dean

IPC: G10L15/00 ,  G06F17/27 ,  G10L15/06

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for computing numeric representations of words. One of the methods includes obtaining a set of training data, wherein the set of training data comprises sequences of words; training a classifier and an embedding function on the set of training data, wherein training the embedding function comprises obtained trained values of the embedding function parameters; processing each word in the vocabulary using the embedding function in accordance with the trained values of the embedding function parameters to generate a respective numerical representation of each word in the vocabulary in the high-dimensional space; and associating each word in the vocabulary with the respective numeric representation of the word in the high-dimensional space.

公开(公告)号：US10169711B1

公开(公告)日：2019-01-01

申请号：US15277306

申请日：2016-09-27

Applicant: Google LLC

Inventor： Jeffrey A. Dean

IPC: G06F7/00 ,  G06F17/30 ,  G06N7/00 ,  H04L29/08 ,  G06F21/62

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for predicting actions based on large-scale aggregations of data. One of the methods includes obtaining user activity data organized into sessions, the user activity data representing user activities, each session including one or more user activities for a particular user, the sessions including sessions for multiple users; receiving a session query, the session query including a query term representing a query activity; identifying matching sessions, the matching sessions each satisfying the session query; identifying likely activities in the matching sessions, likely activities being activities found in the matching sessions that satisfy the session query and occur in the matching sessions more frequently than in sessions in general; and identifying one or more of the likely activities in a response to the session query.

公开(公告)号：US20230325657A1

公开(公告)日：2023-10-12

申请号：US17972466

申请日：2022-10-24

Applicant: Google LLC

Inventor： Gregory S. Corrado ,  Kai Chen ,  Jeffrey A. Dean ,  Gary R. Holt ,  Julian P. Grady ,  Sharat Chikkerur ,  David W. Sculley, II

IPC: G06N3/08 ,  G06N3/084 ,  G06F7/483 ,  G06F17/16 ,  G06N3/045 ,  G06N3/04

CPC classification number: G06N3/08 ,  G06N3/084 ,  G06F7/483 ,  G06F2207/483 ,  G06N3/045 ,  G06N3/04 ,  G06F17/16

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using embedded function with a deep network. One of the methods includes receiving an input comprising a plurality of features, wherein each of the features is of a different feature type; processing each of the features using a respective embedding function to generate one or more numeric values, wherein each of the embedding functions operates independently of each other embedding function, and wherein each of the embedding functions is used for features of a respective feature type; processing the numeric values using a deep network to generate a first alternative representation of the input, wherein the deep network is a machine learning model composed of a plurality of levels of non-linear operations; and processing the first alternative representation of the input using a logistic regression classifier to predict a label for the input.

公开(公告)号：US11687832B1

公开(公告)日：2023-06-27

申请号：US16983979

申请日：2020-08-03

Applicant: Google LLC

Inventor： Gregory S. Corrado ,  Kai Chen ,  Jeffrey A. Dean ,  Samy Bengio ,  Rajat Monga ,  Matthieu Devin

IPC: G06N20/00 ,  G06N3/063 ,  G06N3/08 ,  G06N7/08 ,  G06N5/025 ,  G06F18/214 ,  G06F18/2411 ,  G06N7/01

CPC classification number: G06N20/00 ,  G06N3/063 ,  G06N3/08 ,  G06N7/08 ,  G06F18/214 ,  G06F18/2411 ,  G06N5/025 ,  G06N7/01

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a model using parameter server shards. One of the methods includes receiving, at a parameter server shard configured to maintain values of a disjoint partition of the parameters of the model, a succession of respective requests for parameter values from each of a plurality of replicas of the model; in response to each request, downloading a current value of each requested parameter to the replica from which the request was received; receiving a succession of uploads, each upload including respective delta values for each of the parameters in the partition maintained by the shard; and updating values of the parameters in the partition maintained by the parameter server shard repeatedly based on the uploads of delta values to generate current parameter values.

公开(公告)号：US11055259B2

公开(公告)日：2021-07-06

申请号：US16268149

申请日：2019-02-05

Applicant: Google LLC

Inventor： Yasushi Saito ,  Sanjay Ghemawat ,  Jeffrey A. Dean

IPC: G06F16/16 ,  G06F16/11 ,  G06F16/182 ,  G06F16/215 ,  G06F16/174

Abstract: A method for deleting obsolete files from a file system is provided. The method includes receiving a request to delete a reference to a first target file of a plurality of target files stored in a file system, the first target file having a first target file name. A first reference file whose file name includes the first target file name is identified. The first reference file is deleted from the file system. The method further includes determining whether the file system includes at least one reference file, distinct from the first reference file, whose file name includes the first target file name. In accordance with a determination that the file system does not include the at least one reference file, the first target file is deleted from the file system.

公开(公告)号：US10922488B1

公开(公告)日：2021-02-16

申请号：US16363460

申请日：2019-03-25

Applicant: Google LLC

Inventor： Tomas Mikolov ,  Kai Chen ,  Gregory S. Corrado ,  Jeffrey A. Dean

IPC: G10L15/00 ,  G06F40/279 ,  G10L15/06 ,  G06N20/00 ,  G06F40/30

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for computing numeric representations of words. One of the methods includes obtaining a set of training data, wherein the set of training data comprises sequences of words; training a classifier and an embedding function on the set of training data, wherein training the embedding function comprises obtained trained values of the embedding function parameters; processing each word in the vocabulary using the embedding function in accordance with the trained values of the embedding function parameters to generate a respective numerical representation of each word in the vocabulary in the high-dimensional space; and associating each word in the vocabulary with the respective numeric representation of the word in the high-dimensional space.

公开(公告)号：US10740301B2

公开(公告)日：2020-08-11

申请号：US15868928

申请日：2018-01-11

Applicant: Google LLC

Inventor： Jeffrey A. Dean ,  Sanjay Ghemawat ,  Andrew B. Fikes ,  Yasushi Saito

IPC: G06F16/182 ,  G06F16/22 ,  G06F9/50 ,  G06F16/13 ,  H04L29/08

Abstract: A method of accessing data includes storing a table that includes a plurality of tablets corresponding to distinct non-overlapping table portions. Respective pluralities of tablet access objects and application objects are stored in a plurality of servers. A distinct application object and distinct tablet are associated with each tablet access object. Each application object corresponds to a distinct instantiation of an application associated with the table. The tablet access objects and associated application objects are redistributed among the servers in accordance with a first load-balancing criterion. A first request directed to a respective tablet is received from a client. In response, the tablet access object associated with the respective tablet is used to perform a data access operation on the respective tablet, and the application object associated with the respective tablet is used to perform an additional computational operation to produce a result to be returned to the client.