公开(公告)号：US11270059B2

公开(公告)日：2022-03-08

申请号：US16552210

申请日：2019-08-27

Applicant: Microsoft Technology Licensing, LLC

Inventor： Ji Li ,  Xiaozhi Yu ,  Gregory Alexander DePaul ,  Youjun Liu ,  Amit Srivastava

IPC: G06F40/106 ,  G06N20/00

Abstract: A textual user input is received and a plurality of different text-to-content models are run on the textual user input. A selection system attempts to identify a suggested content item, based upon the outputs of the text-to-content models. The selection system first attempts to generate a completed suggestion based on outputs from a single text-to-content model. It then attempts to mix the outputs of the text-to-content models to obtain a completed content suggestion.

公开(公告)号：US12001514B2

公开(公告)日：2024-06-04

申请号：US18047324

申请日：2022-10-18

Applicant: Microsoft Technology Licensing, LLC

Inventor： Ji Li ,  Youjun Liu ,  Amit Srivastava

IPC: G06F18/21 ,  G06F18/25 ,  G06F21/62 ,  G06N20/00

CPC classification number: G06F18/217 ,  G06F18/254 ,  G06F21/6218 ,  G06N20/00

Abstract: The present disclosure relates to processing operations that execute image classification training for domain-specific traffic, where training operations are entirely compliant with data privacy regulations and policies. Image classification model training, as described herein, is configured to classify meaningful image categories in domain-specific scenarios where there is unknown data traffic and strict data compliance requirements that result in privacy-limited image data sets. Iterative image classification training satisfies data compliance requirements through a combination of online image classification training and offline image classification training. This results in tuned image recognition classifiers that have improved accuracy and efficiency over general image recognition classifiers when working with domain-specific data traffic. One or more image recognition classifiers are independently trained and tuned to detect an image class for image classification. Training of independent image recognition classifiers is also utilized for training and tuning of deeper learning models for image classification.

公开(公告)号：US20200265153A1

公开(公告)日：2020-08-20

申请号：US16276908

申请日：2019-02-15

Applicant: Microsoft Technology Licensing, LLC

Inventor： Ji Li ,  Youjun Liu ,  Amit Srivastava

IPC: G06F21/62 ,  G06K9/62 ,  G06N20/00

Abstract: The present disclosure relates to processing operations that execute image classification training for domain-specific traffic, where training operations are entirely compliant with data privacy regulations and policies. Image classification model training, as described herein, is configured to classify meaningful image categories in domain-specific scenarios where there is unknown data traffic and strict data compliance requirements that result in privacy-limited image data sets. Iterative image classification training satisfies data compliance requirements through a combination of online image classification training and offline image classification training. This results in tuned image recognition classifiers that have improved accuracy and efficiency over general image recognition classifiers when working with domain-specific data traffic. One or more image recognition classifiers are independently trained and tuned to detect an image class for image classification. Training of independent image recognition classifiers is also utilized for training and tuning of deeper learning models for image classification.

公开(公告)号：US20210064690A1

公开(公告)日：2021-03-04

申请号：US16552210

申请日：2019-08-27

Applicant: Microsoft Technology Licensing, LLC

Inventor： Ji Li ,  Xiaozhi Yu ,  Gregory Alexander DePaul ,  Youjun Liu ,  Amit Srivastava

IPC: G06F17/21 ,  G06N20/00

Abstract: A textual user input is received and a plurality of different text-to-content models are run on the textual user input. A selection system attempts to identify a suggested content item, based upon the outputs of the text-to-content models. The selection system first attempts to generate a completed suggestion based on outputs from a single text-to-content model. It then attempts to mix the outputs of the text-to-content models to obtain a completed content suggestion.

公开(公告)号：US11507677B2

公开(公告)日：2022-11-22

申请号：US16276908

申请日：2019-02-15

Applicant: Microsoft Technology Licensing, LLC

Inventor： Ji Li ,  Youjun Liu ,  Amit Srivastava

IPC: G06F21/62 ,  G06N20/00 ,  G06K9/62

Abstract: The present disclosure relates to processing operations that execute image classification training for domain-specific traffic, where training operations are entirely compliant with data privacy regulations and policies. Image classification model training, as described herein, is configured to classify meaningful image categories in domain-specific scenarios where there is unknown data traffic and strict data compliance requirements that result in privacy-limited image data sets. Iterative image classification training satisfies data compliance requirements through a combination of online image classification training and offline image classification training. This results in tuned image recognition classifiers that have improved accuracy and efficiency over general image recognition classifiers when working with domain-specific data traffic. One or more image recognition classifiers are independently trained and tuned to detect an image class for image classification. Training of independent image recognition classifiers is also utilized for training and tuning of deeper learning models for image classification.

公开(公告)号：US10891514B2

公开(公告)日：2021-01-12

申请号：US16222905

申请日：2018-12-17

Applicant: Microsoft Technology Licensing, LLC

Inventor： Youjun Liu ,  Ji Li ,  Amit Srivastava

IPC: G06K9/62 ,  G06K9/00 ,  G06N3/08

Abstract: The present disclosure relates to processing operations configured for an image recognition pipeline that is used to tailor real-time management of image recognition processing for technical scenarios across a plurality of different applications/services. Image recognition processing is optimized at run-time to ensure that latency requirements are met so that image recognition processing results are returned in a timely manner that aids task execution in an application-specific instances. An image recognition pipeline may manage a plurality of image recognition models that comprise a combination of image analysis service (IAS) models and deep learning models. A scheduler of the image recognition pipeline optimizes image recognition processing by selecting at least: a subset of the image recognition models for image recognition processing and a device configuration for execution of the subset of image recognition models, in order to return image recognition results within a threshold time period that satisfies application-specific execution.