公开(公告)号：WO2022005616A1

公开(公告)日：2022-01-06

申请号：PCT/US2021/031472

申请日：2021-05-10

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： LIU, Yunxin ,  HOU, Jiahui

IPC: G06F21/62 ,  G06F21/74 ,  G06F21/53 ,  G06N3/02 ,  G06F21/6218 ,  G06N3/0454 ,  G06N3/0472 ,  G06N3/063

Abstract: According to implementations of the subject matter described herein, there is provided a solution for secure execution of a machine learning network. An operation of a first network layer of a machine learning network is executed in an uTEE of a computing device based on an input of the first network layer and a first set of modified parameter values, to obtain a first error intermediate output. The modified parameter values are determined by modifying at least one subset of parameter values of the first network layer with first secret data. A first corrected intermediate output is determined in a TEE of the computing device by modifying the first error intermediate output at least based on the input and first secret data. A network output is determined based on the first corrected intermediate output. In this way, it is possible to protect the confidentiality of the machine learning network.

公开(公告)号：WO2017040552A1

公开(公告)日：2017-03-09

申请号：PCT/US2016/049537

申请日：2016-08-31

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： LUO, Qi ,  LIU, Yunxin ,  ZHANG, Sam ,  TAN, Kun ,  YIN, Qiufeng ,  ZHANG, Lintao ,  ZHAO, Feng

IPC: H04L29/06 ,  H04W12/06 ,  H04L29/08

CPC classification number: H04L63/08 ,  H04L67/025 ,  H04W12/06

Abstract: Various systems and methods for relaying remote a request are described herein. In one example, a method includes receiving a request at a public website to access a private router. The method can also include authenticating the request via an authentication service. Furthermore, the method can include providing access to the private router via the public website upon authentication.

Abstract translation: 这里描述了用于中继远程请求的各种系统和方法。 在一个示例中，一种方法包括在公共网站接收访问专用路由器的请求。 该方法还可以包括通过认证服务来认证请求。 此外，该方法可以包括在认证时通过公共网站提供对私有路由器的访问。

公开(公告)号：WO2023272662A1

公开(公告)日：2023-01-05

申请号：PCT/CN2021/103872

申请日：2021-06-30

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： JIANG, Shiqi ,  LI, Yuanchun ,  SHU, Yuanchao ,  LIU, Yunxin

IPC: G06T7/20

Abstract: Implementations of the present disclosure provide a solution for object detection. In this solution, object distribution information and performance metrics are obtained. The object distribution information indicates a size distribution of detected objects in a set of historical images captured by a camera. The performance metric indicates corresponding performance levels of a set of predetermined object detection models. At least one detection plan is further generated based on the object distribution information and the performance metric. The at least one detection plan indicates which of the set of predetermined object detection models is to be applied to each of at least one sub-image in a target image to be captured by the camera. Additionally, the at least one detection plan is provided for object detection on the target image. In this way, a balance between the detection latency and the detection accuracy may be improved.

公开(公告)号：WO2020256836A1

公开(公告)日：2020-12-24

申请号：PCT/US2020/030327

申请日：2020-04-29

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： ZHANG, Chen ,  LIU, Yunxin

IPC: G06N3/04 ,  G06N3/10

Abstract: Various implementations of the subject matter as described herein relate to a sparse convolutional neural network. In some implementations, a computer-implemented method comprises: quantizing an input feature map to obtain a quantized input feature map; determining, based on the quantized input feature map, a sparsity mask for an output feature map through a quantized version of a convolutional neural network, the sparsity mask indicating positions of non-zero entries in the output feature map; and determining, based on the input feature map, the non-zero entries indicated by the sparsity mask in the output feature map through the convolutional neural network.

公开(公告)号：WO2020242689A1

公开(公告)日：2020-12-03

申请号：PCT/US2020/030018

申请日：2020-04-27

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： LIU, Yunxin ,  ZHANG, Lintao

IPC: G06N3/04 ,  G06F21/50

Abstract: In accordance with implementations of the subject matter described herein, there is provided a solution for execution of a deep learning model. In the solution, partitioned convolutions are executed based on an input and a set of parameter values of the convolutional layer sequentially in a trusted execution environment (TEE) of a computing device. The execution of a given one of partitioned convolutions comprises: storing, into a protected memory area in the TEE, an input portion of the input to be processed by a subset of parameter values for the given partitioned convolution; determining a result of the given partitioned convolution through a single matrix multiplication operation; and removing the input portion. By combining results of the partitioned convolutions, a result of the convolution is determined. Therefore, the solution can accelerate the model execution speed and improve the storage efficiency in a highly safe TEE with limited memory resources.

公开(公告)号：WO2016210206A1

公开(公告)日：2016-12-29

申请号：PCT/US2016/039133

申请日：2016-06-24

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： LIU, Yunxin ,  ZHOU, Hucheng

IPC: G09G5/36

CPC classification number: G09G5/363 ,  G09G5/373 ,  G09G2340/045 ,  G09G2354/00 ,  G09G2360/08

Abstract: A computing device may dynamically adjust a pixel density based at least in part on a viewing distance between a user and a display of the computing device. In some examples, the viewing distance may be determined using low power acoustic (e.g., ultrasonic) sensing. A pixel density at which to display content may be determined using algorithms based on the viewing distance and a visual acuity of a user. Content to be displayed on the computing device may be sent to processors of the computing device for graphics processing. In some examples, the content may be intercepted, such as by using a hooking process, before processing and scaled based on the determined pixel density. Scaling down the pixel density of the content may require less system resources to process the content, which may result in less power consumption by the processors to perform the graphics processing operations.

Abstract translation: 计算设备可以至少部分地基于用户和计算设备的显示器之间的观看距离来动态地调整像素密度。 在一些示例中，可以使用低功率声学（例如，超声波）感测来确定观看距离。 可以使用基于观看距离和用户的视敏度的算法来确定显示内容的像素密度。 要显示在计算设备上的内容可以被发送到用于图形处理的计算设备的处理器。 在一些示例中，可以例如通过使用挂钩过程在处理之前拦截内容，并基于所确定的像素密度进行缩放。 缩小内容的像素密度可能需要较少的系统资源来处理内容，这可能导致处理器执行图形处理操作的功耗较少。

公开(公告)号：WO2019040187A1

公开(公告)日：2019-02-28

申请号：PCT/US2018/040666

申请日：2018-07-03

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： LIU, Yunxin ,  ZHANG, Jiansong ,  ZHANG, Lintao ,  MOSCIBRODA, Thomas

IPC: G09G5/36 ,  G09G5/397 ,  H03M7/40

Abstract: Embodiments of the subject matter described herein relate to a wireless programmable media processing system. In the media processing system, a processing unit in a computing device generates a frame to be displayed based on a graphics content for an application running on the computing device. The frame to be displayed is then divided into a plurality of block groups which are compressed. The plurality of compressed block groups are sent to a graphics display device over a wireless link. In this manner, both the generation and the compression of the frame to be displayed may be completed at the same processing unit in the computing device, which avoids data copying and simplifies processing operations. Thereby, the data processing speed and efficiency is improved significantly.

公开(公告)号：WO2017039882A1

公开(公告)日：2017-03-09

申请号：PCT/US2016/044147

申请日：2016-07-27

Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor： LIU, Yunxin ,  YIN, Qiufeng ,  ZHANG, Lintao ,  ZHAO, Feng ,  WU, Ming ,  LUO, Qi ,  ZHANG, Sam

IPC: H04L12/701

CPC classification number: H04L45/306 ,  H04L41/5054 ,  H04L43/10 ,  H04L45/74

Abstract: Various systems and methods for providing network services are described herein. In one example, a method includes receiving, via a first processor, a network packet from a source device. The method can also include sending, via the first processor, the network packet to a second processor if a service subsystem and a service are responsive. Furthermore, the method can include modifying, via the second processor, the network packet based on the service. The method can also include receiving, via the first processor, the modified network packet from the second processor. The method can also further include sending, via the first processor, the network packet to a destination device.

Abstract translation: 本文描述了用于提供网络服务的各种系统和方法。 在一个示例中，一种方法包括经由第一处理器从源设备接收网络分组。 该方法还可以包括：如果服务子系统和服务是响应的，则经由第一处理器将网络分组发送到第二处理器。 此外，该方法可以包括经由第二处理器修改基于该服务的网络分组。 该方法还可以包括经由第一处理器从第二处理器接收经修改的网络分组。 该方法还可以进一步包括经由第一处理器将网络分组发送到目的地设备。