Multi-task machine learning for predicted touch interpretations

    公开(公告)号:US10261685B2

    公开(公告)日:2019-04-16

    申请号:US15393611

    申请日:2016-12-29

    Applicant: Google Inc.

    Abstract: The present disclosure provides systems and methods that leverage machine learning to predict multiple touch interpretations. In particular, the systems and methods of the present disclosure can include and use a machine-learned touch interpretation prediction model that has been trained to receive touch sensor data indicative of one or more locations of one or more user input objects relative to a touch sensor at one or more times and, in response to receipt of the touch sensor data, provide one or more predicted touch interpretation outputs. Each predicted touch interpretation output corresponds to a different type of predicted touch interpretation based at least in part on the touch sensor data. Predicted touch interpretations can include a set of touch point interpretations, a gesture interpretation, and/or a touch prediction vector for one or more future times.

    Systems and methods of sourcing hours of operation for a location entity

    公开(公告)号:US10257660B2

    公开(公告)日:2019-04-09

    申请号:US14978252

    申请日:2015-12-22

    Applicant: Google Inc.

    Abstract: Computer-implemented methods and systems for sourcing hours of operation for a business or other location entity can include receiving a request to update hours of operation. A number of mobile devices present at the location entity during one or more periods of time can be determined based at least in part on a collection of time-stamped mobile device location data that identifies mobile devices associated with discrete users present at the location entity. The number of mobile devices and initial operating hours for the location entity can be provided as input to a statistical model (e.g., neural network, support-vector machine (SVM) or logistic regression model). An output of the model can indicate a likelihood that operating hours for the location entity have changed relative to the initial operating hours. The operating hours for the location entity can then be updated based at least in part on the model output.

    Generating a composite image from a physical item

    公开(公告)号:US10257485B2

    公开(公告)日:2019-04-09

    申请号:US15350840

    申请日:2016-11-14

    Applicant: Google Inc.

    Abstract: A computer-implemented method includes capturing, with a camera, a first image of a physical item at a first camera position, detecting borders associated with the physical item, based on the first image, generating an overlay that includes a plurality of objects that are positioned within one or more of the borders associated with the physical item, capturing, with the camera, subsequent images of the physical item, where each subsequent image is captured with a respective subsequent camera position, and during capture of the subsequent images, displaying an image preview that includes the overlay. The method further includes establishing correspondence between pixels of the first image and pixels of each of the subsequent images and generating a composite image of the physical item, where each pixel value of the composite image is based on corresponding pixel values of the first image and the subsequent images.

    IOT interaction system
    176.
    发明授权

    公开(公告)号:US10257256B2

    公开(公告)日:2019-04-09

    申请号:US15001813

    申请日:2016-01-20

    Applicant: Google Inc.

    Abstract: Systems and methods for facilitating interactions with embedded devices are provided. In one embodiment, a method can include obtaining a first set of data indicative of at least a plurality of interactions between a user device and a plurality of embedded devices associated with a building, and one or more locations of the user device associated with each respective interaction. The method can include generating a second set of data for each embedded device based, at least in part, on the first set of data. Each second set of data can be indicative of at least a number of interactions between the user device and the respective embedded device for each location. The method can include determining a particular location of the user device. The method can include identifying one or more recommended embedded devices and providing information about at least one recommended embedded device to the user device.

    Annotating a transaction history record with merchant information identified from a merchant identifier and user computing device location data

    公开(公告)号:US10255631B2

    公开(公告)日:2019-04-09

    申请号:US14504148

    申请日:2014-10-01

    Applicant: GOOGLE INC.

    Abstract: A payment processing system receives one or more indications of a user's location at a time that the user initiates a transaction with a merchant system at a physical location of the merchant system, processes the transaction, and receives a merchant point of sale identifier from the issuer system associated with the financial account information selected by the user. The payment processing system creates a combined list of candidate merchant identities based on the received merchant POS identifier and the user computing device location. The payment processing system calculates a probability for each of the merchant identities in the list based on similarity to the merchant POS identifier and user location data. The payment processing system associates the merchant POS identifier with the merchant identity having the highest probability and creates an annotated transaction record to display to the user.

    Document object model data extraction using HTML tag identifier

    公开(公告)号:US10255254B1

    公开(公告)日:2019-04-09

    申请号:US15091363

    申请日:2016-04-05

    Applicant: Google Inc.

    Inventor: Jeffrey Getzin

    Abstract: Document object model (DOM) extraction using a hypertext markup language (HTML) tag identifier is described. A web browser executing on a computing device receives content preference data input via a content preferences menu of the web browser. The web browser adds, to the DOM of an electronic resource, the HTML tag identifier that indicates the content preference data. The web browser renders the electronic resource with the HTML tag identifier. A content server parses the electronic resource rendered by the web browser to identify the HTML tag identifier, and selects the content item corresponding the content preference data indicated by the HTML tag identifier. The web browser receives the content item and renders the content item for display on a display device of the computing device.

    Electronic device with modulated light flash operation for rolling shutter image sensor

    公开(公告)号:US10250789B2

    公开(公告)日:2019-04-02

    申请号:US15426675

    申请日:2017-02-07

    Applicant: Google Inc.

    Inventor: Johnny C Lee

    Abstract: An electronic device (100) includes a modulated light projector (119) and an electronic rolling shutter (ERS) imaging camera (1302) having a sensor array (1306) with a set of pixel rows. The electronic device includes a controller (1304) to control the modulated light projector (119) to project a modulated light flash (1608, 1810, 2010) during capture of a first image frame (1606, 1806, 2006) by the sensor array. The controller is to initiate the modulated light flash while each pixel row of the set is exposed for gathering light and to terminate the modulated light flash before a pixel row of the set is exposed for capture of a next image frame by the sensor array. The electronic device further includes a processor (802) to determine modulated light image data (1320) based on the first image frame. The controller also may control the modulated light projector to refrain from projecting the modulated light flash during capture of a second image frame (1602, 1612, 1802, 1816, 2002, 2018), adjacent to the first image frame.

Patent Agency Ranking