Microsoft Technology Licensing, LLC

Webtv Networks, Inc.

Microsoft Corporation

Christopher M White

Phillip York Goldman

David R Anderson

Darren Keith Edge

Bruce A Leak

Jonathan Karl Larson

Keith R Ohlfs

Timo A Bruck

Lee S Mighdoll

Nathan Roy Evans

Jeffrey D Yaksick

Weiwei Yang

Kateryna Lytvynets

Stephen G Perlman

David Hendler Sloo

Daniel John Zigmond

Johnson T Apacible

Joe Freeman Britt, Jr

Sean P Nolan

Daniel Danker

John R Matheny

Mark H Krueger

Jeffrey Dick Jones

Brian J Guarraci

Vijay Varadan

Anson Tsao

Patrick P Bonnaure

Robert Andrew Little

Zeke Koch

Tjeerd Hoek

Eric B Fox

Niels Thomas Ferguson

Richard M Banks

David Geoffrey Molyneaux

Oriana Riva

Glenn L Frankel

David Charles LeBlanc

Thomas Scott Coon

Michael Ronn Marcelais

Benjamin J Bunker

Prashant Dwarkadas Agrawal

Ambrose Thomas Treacy

Amber Dawn Hoak

Maithili Dandige

Lawrence A Kenyon

Aaron Weiss

Catherine Rose Morrow

Aaron Everett Erlandson

Kushal Arora

Muzammil Ahmed

Pratyay Banerjee

Sahisnu Mazumder

Shweti Mahajan

Steve Waserman

Dave Molyneaux

A computer implemented method includes accessing training data that includes images, associated alternative text, and proximately located text providing instructions describing a user interface. The images are paired with text captions derived from the proximately located text and image's alternative text, training a vision and language model in a self-supervised manner using language masking, region masking, and image-text alignment techniques on respective image region features and tokenized text captions. Fine-tuning of the vision and language model is performed to obtain a specialized model representing user interface elements and associated functions.

Self-supervised system for learning a user interface language

Systems and methods for modeling higher-level metrics from graph data derived from already-collected but not yet connected data are disclosed. A method includes extracting a first set of actor-related data, a second set of object-related data, and a third set of temporal data from the set of the already-collected but not yet connected data representative of a unit-level contribution to the target activity. The method further includes generating graph data for a graph using the set of the already-collected but not yet connected data, where each of the plurality of nodes of the graph corresponds to the actor or the object, and where an attribute associated with each of the plurality of edges of the graph corresponds to a measurement associated with the target activity. The method further includes modeling a relationship between graph attributes associated with the graph data and a higher-level metric associated with the target activity.

Modeling higher-level metrics from graph data derived from already-collected but not yet connected data

Systems and methods for graph embedding already-collected but not yet connected data are disclosed. A method includes extracting a first set of actor-related data, a second set of object-related data, and a third set of temporal data from a set of the already-collected but not yet connected data representative of a unit-level contribution to the target activity. The method further includes generating graph data for at least one graph having a plurality of nodes and a plurality of edges using the set of the already-collected but not yet connected data, where each of the plurality of nodes corresponds to the actor or the object, and where an attribute associated with each of the plurality of edges corresponds to a measurement associated with the target activity during a temporal dimension of interest. The method further includes converting the graph data into metric space data using a graph embedding process.

Graph embedding already-collected but not yet connected data

Systems and methods for managing content provenance are provided. A network system accesses a plurality of documents. The plurality of documents is then hashed to identify one or more content features within each of the documents. In one embodiment, the hash is a MinHash. The network system compares the content features of each of the plurality of documents to determine a similarity score between each of the plurality of documents. In one embodiment, the similarly score is a Jaccard score. The network system then clusters the plurality of documents into one or more clusters based on the similarity score of each of the plurality of documents. In one embodiment, the clustering is performed using DBSCAN. DBSCAN can be iteratively performed with decreasing epsilon values to derive clusters of related but relatively dissimilar documents. The clustering information associated with the clusters are stored for use during runtime.

Revealing content reuse using coarse analysis

A set of unstructured data is analyzed to infer structural elements from the unstructured data, and quantized data quality levels, indicative of data quality in the structural elements, are assigned to the structural elements. A set of structured data is generated to include the structural elements inferred from the unstructured data and associations between respective ones of the structural elements in the set of structured data and the corresponding quantized quality levels assigned to the structural elements. The set of structured data, including the associations between respective ones of the structural elements and the corresponding quantized quality levels assigned to the structural elements, is provided to a user interface application to enable the user interface application to visually display varying data qualities in the set of structured data.

Quality-aware data interfaces

Techniques for synthesizing and analyzing data are disclosed. A ML model anonymizes microdata to generate synthesized data. This anonymizing is performed by reproducing attributes identified within microdata and by applying constraints to prevent rare attribute combinations from being reproduced in the synthesized data. User input selects attributes to filter the synthesized data, thereby generating a subset of records. A UI displays a synthesized aggregate count representing how many records are in the subset. Pre-computed aggregate counts are accessed to indicate how many records in the microdata embody certain attributes. Based on the user input, there is an attempt to identify a particular count from the pre-computed aggregate counts. This count reflects how many records of the microdata would remain if the selected attributes were used to filter the microdata. That count is displayed along with the synthesized aggregate count. The two counts are juxtaposed next to one another.

Privacy-preserving data platform

Systems and methods for managing content provenance are provided. A network system accesses a document of a plurality of documents to be analyzed. The network system extracts text fragments from the document including a first fragment and a second fragment. A determination is made whether each of the text fragments match an entry in a hash table. Based on a first fragment not matching any entries in the hash table, the network system creates a new entry in the hash table, whereby the first fragment is used to generate a key in the hash table. Based on a second fragment matching an entry of the hash table, the network system associates the document with a key of the matching entry in the hash table, whereby the associating comprising updating the hash table with an identifier of the document.

Revealing content reuse using fine analysis

A system and method for managing operation of electrical devices includes a control module that monitors status of multiple sources of electrical power to one or more electrical devices and electrical usage of the one or more electrical devices that receive electricity from the source of electrical power. The operation of the one or more electrical devices is managed using a machine learning model that forecasts status of the at least one source of electrical power and generates operational rules for the one or more electrical devices from historical values of control parameters of the one or more electrical devices, the status of the source of electrical power, and the electrical usage of the one or more electrical devices. The system may optimize renewable energy utilization, power grid stabilization, cost of electrical power usage, and the like.

Power grid aware machine learning device

A system and method for regulating charge/discharge of a battery to stabilize a regional power grid includes a regulation control module that monitors a frequency regulation signal from the regional power grid and market conditions for obtaining power from the regional power grid. A machine learning module predicts from the frequency regulation signal and market conditions a future beneficial period when the battery may be exposed to the regional power grid to charge/discharge power in accordance with the frequency regulation signal to stabilize the regional power grid through participation in the regulation of the regional power grid. Another machine learning module calculates a regulation control signal that tracks the frequency regulation signal during the future beneficial period and outputs the regulation control signal to at least one battery control module that manages charging/discharging of the battery to selectively withdraw/apply power from/to the regional power grid during the future beneficial period.

Datacenter stabilization of regional power grids

Embodiments are disclosed that relate to tracking one or more objects during a process that utilizes the objects. For example, one embodiment provides a method for monitoring performance of a process involving one or more objects, wherein the method includes receiving a set of rules defining one or more portions of the process and receiving object identification information regarding the one or more objects. The method further includes, for a selected portion of the process, receiving image information of a physical scene, identifying from the image information and the object identification information an operation performed with an identified object in the physical scene, and taking an action based upon whether the operation satisfies a rule of the set of rules associated with the selected portion of the process.

Tracking objects during processes

Embodiments are disclosed that relate to producing a synthetic environmental model derived from a three dimensional representation of an environment, and rendering images from the model. For example, one disclosed embodiment provides a method including detecting a trigger to build the synthetic environmental model utilizing the three dimensional representation of the environment, and, in response to the trigger, obtaining a set of synthetic image elements for use in constructing the synthetic environmental model. The method further includes fitting one or more elements from the set of synthetic image elements to the three dimensional representation of the environment according to a set of rules to produce the synthetic environmental model, and rendering an image from the synthetic environmental model for display, the image showing the one or more elements from the set of synthetic image elements replacing real-world topography in the environment.

Construction of synthetic augmented reality environment

A consent management system is described which manages an entity's consent to consume application functionality on a per-feature level of granularity. To perform this task, the consent management system maintains consent information which describes a plurality of use conditions associated with the features of the application functionality. In one case, the consent information has a hierarchical data structure that identifies a hierarchy of application features. In operation, the consent management system accesses the content information whenever an entity seeks to access a particular application feature; it grants or denies access to the feature based on the consent information. The consent management system also includes a consent maintenance module that allows an administrator (or other entity) to cancel or modify any use condition in the consent information.

Progressive consent

An authorization framework is provided that protects data records in a platform, such as a service-based platform, by requiring multiple level entities to be authorized with respect to the data records. For example, the data records can have an associated owner user that can grant access to other users with respect to the data. Additionally, however, the user can also grant access to certain applications that access the platform such that the data records can be initially closed for a user requiring the user to explicitly grant desired access to applications and/or users. In this regard, applications can be forbidden from accessing the data, even on behalf of the user, unless expressly authorized to do so by the user. Thus, the user can make informed decisions regarding who is to have access to its data.

Multiple entity authorization model

A system for accessing the World Wide Web or other information sources using the Internet incorporates scaled-down versions of images associated with recently-viewed documents or documents designated as favorites into a recent menu or a favorites menu, respectively. As a user accesses a Web page, a scaled-down version of the Web page is inserted into a recent menu. The user can view the recent menu and visually determine the identity of the recently-viewed Web pages, which allows the user to conveniently select Web pages to access. This recent menu also enables the user to access recently-viewed documents in any order, which does not need to be in the chronological order in which the pages were originally viewed or in a reverse chronological order. Scaled-down versions of Web pages designated as favorites are stored in the favorites menu, thereby assisting the user in recalling the content of the favorite documents.

Displaying scaled-down versions of documents on favorites and recent menu

Growing community of inventors

San Francisco, CA, United States of America

Christopher M White