公开(公告)号：US20220004918A1

公开(公告)日：2022-01-06

申请号：US16946779

申请日：2020-07-06

Applicant: Google LLC

Inventor： Spurthi Amba Hombaiah ,  Vladimir Ofitserov ,  Mike Bendersky ,  Marc Alexander Najork

IPC: G06N20/00 ,  G06N5/04 ,  G06F16/9038

Abstract: Implementations relate to training a model that can be used to process values for defined features, where the values are specific to a user account, to generate a predicted user measure that reflects both popularity and quality of the user account. The model is trained based on losses that are each generated as a function of both a corresponding generated popularity measure and a corresponding generated quality measure of a corresponding training instance. Accordingly, the model can be trained to generate, based on values for a given user account, a single measure that reflects both quality and popularity of the given user account. Implementations are additionally or alternatively directed to utilizing such predicted user measures to restrict provisioning of content items that are from user accounts having respective predicted user measures that fail to satisfy a threshold.

公开(公告)号：US20230222285A1

公开(公告)日：2023-07-13

申请号：US17928984

申请日：2020-12-22

Applicant: Google LLC

Inventor： Mingyang Zhang ,  Cheng Li ,  Tao Chen ,  Spurthi Amba Hombaiah ,  Michael Bendersky ,  Marc Alexander Najork ,  Te-Lin Wu

IPC: G06F40/166 ,  G06F40/284 ,  G06V30/413 ,  G06F40/109

CPC classification number: G06F40/166 ,  G06F40/284 ,  G06V30/413 ,  G06F40/109

Abstract: Systems and methods for document processing that can process and understand the layout, text size, text style, and multimedia of a document can generate more accurate and informed document representations. The layout of a document paired with text size and style can indicate what portions of a document are possibly more important, and the understanding of that importance can help with understanding of the document. Systems and methods utilizing a hierarchical framework that processes the block-level and the document-level of a document can capitalize on these indicators to generate a better document representation.

公开(公告)号：US20230094198A1

公开(公告)日：2023-03-30

申请号：US18074774

申请日：2022-12-05

Applicant: GOOGLE LLC

Inventor： Spurthi Amba Hombaiah ,  Vladimir Ofitserov ,  Mike Bendersky ,  Marc Alexander Najork

IPC: G06N20/00 ,  G06F16/9038 ,  G06N5/04

Abstract: Implementations relate to training a model that can be used to process values for defined features, where the values are specific to a user account, to generate a predicted user measure that reflects both popularity and quality of the user account. The model is trained based on losses that are each generated as a function of both a corresponding generated popularity measure and a corresponding generated quality measure of a corresponding training instance. Accordingly, the model can be trained to generate, based on values for a given user account, a single measure that reflects both quality and popularity of the given user account. Implementations are additionally or alternatively directed to utilizing such predicted user measures to restrict provisioning of content items that are from user accounts having respective predicted user measures that fail to satisfy a threshold.

公开(公告)号：US11551150B2

公开(公告)日：2023-01-10

申请号：US16946779

申请日：2020-07-06

Applicant: Google LLC

Inventor： Spurthi Amba Hombaiah ,  Vladimir Ofitserov ,  Mike Bendersky ,  Marc Alexander Najork

IPC: G06N20/00 ,  G06F16/9038 ,  G06N5/04

Abstract: Implementations relate to training a model that can be used to process values for defined features, where the values are specific to a user account, to generate a predicted user measure that reflects both popularity and quality of the user account. The model is trained based on losses that are each generated as a function of both a corresponding generated popularity measure and a corresponding generated quality measure of a corresponding training instance. Accordingly, the model can be trained to generate, based on values for a given user account, a single measure that reflects both quality and popularity of the given user account. Implementations are additionally or alternatively directed to utilizing such predicted user measures to restrict provisioning of content items that are from user accounts having respective predicted user measures that fail to satisfy a threshold.

公开(公告)号：US12236322B2

公开(公告)日：2025-02-25

申请号：US18074774

申请日：2022-12-05

Applicant: GOOGLE LLC

Inventor： Spurthi Amba Hombaiah ,  Vladimir Ofitserov ,  Mike Bendersky ,  Marc Alexander Najork

IPC: G06N20/00 ,  G06F16/9038 ,  G06N5/04

Abstract: Implementations relate to training a model that can be used to process values for defined features, where the values are specific to a user account, to generate a predicted user measure that reflects both popularity and quality of the user account. The model is trained based on losses that are each generated as a function of both a corresponding generated popularity measure and a corresponding generated quality measure of a corresponding training instance. Accordingly, the model can be trained to generate, based on values for a given user account, a single measure that reflects both quality and popularity of the given user account. Implementations are additionally or alternatively directed to utilizing such predicted user measures to restrict provisioning of content items that are from user accounts having respective predicted user measures that fail to satisfy a threshold.

公开(公告)号：US20230401382A1

公开(公告)日：2023-12-14

申请号：US18249275

申请日：2021-10-19

Applicant: Google LLC

Inventor： Spurthi Amba Hombaiah ,  Mingyang Zhang ,  Michael Bendersky ,  Tao Chen ,  Marc Alexander Najork

IPC: G06F40/242 ,  G06F40/40 ,  G06F40/30 ,  G06F40/284

CPC classification number: G06F40/242 ,  G06F40/40 ,  G06F40/30 ,  G06F40/284

Abstract: Provided are systems and methods for incremental training of machine learning models to adapt to changes in an underlying data distribution. One example setting in which the techniques described herein may be beneficial is for incrementally training natural language models to enable the models to have or adapt to a dynamically changing vocabulary. Incremental training is provided as a feasible and inexpensive way of adapting machine learning models to evolving vocabulary without having to retrain them from scratch.