Microsoft Technology Licensing, LLC

Dropbox, Inc.

Micosoft Corporation

Geoff Hulten

John Scarrow

Kristofer Noel Iverson

Ivan Osipkov

Ashok Pancily Poothiyot

Anthony Penta

Tim Gasser

Christopher A Meek

Andrew J Newman

Ron Franczyk

Ali Zafar

Stephen Voorhees

Zhengyou Zhang

Zicheng Liu

Michael John Conrad

Xin Huang

Kyle James Krum

Tsung-Hsiang Chang

Ryan Charles Colvin

Robert L McCann

Harry Simon Katz

Chad W Mills

Joshua Korb

Javier Hernandez Rivera

Vasanth Krishna Namasivayam

Eriola Kruja

Steve Rehfuss

Sateesh Srinivasan

Ameya Bhatawdekar

David DeBarr

Paul Steve Rehfuss

The present disclosure relates to systems, non-transitory computer-readable media, and methods for selecting machine-learning models and hardware environments for executing a task. In particular, in one or more embodiments, the disclosed systems select a designated machine-learning model for executing a task based on workload features of the task and task routing metrics for a plurality of machine-learning models. In addition, in one or more embodiments, the disclosed systems select a designated hardware environment for executing the task based on workload features for the task and task routing metrics for a plurality of hardware environments. In some embodiments, the disclosed systems select a fallback machine-learning model and a fallback hardware environment for executing the task if the designated machine-learning model or designated hardware environment are unavailable. Moreover, in one or more embodiments, the disclosed systems can pause and initiate tasks based on bandwidth availability.

Dynamically selecting artificial intelligence models and hardware environments to execute tasks

The present disclosure relates to systems, non-transitory computer-readable media, and methods for generating personal responses through retrieval-augmented generation. In particular, the disclosed systems can generate a query embedding from a query generated by an entity and determine data context specific to the entity by comparing the query embedding with a plurality of vectorized segments of content items associated with the entity. The disclosed systems can provide the data context to a large language model and generate a personalized response informed by the data context. Subsequently, the disclosed systems can provide the personalized response for display on a client device associated with the entity.

Personalized retrieval-augmented generation system

A method of safe file transmission and reputation lookup is provided. As a part of the safe file transmission and reputation lookup methodology, a data file that is to be made available to a data file receiver is accessed and it is determined whether the data file needs to be provided a protective file. The data file is wrapped in a protective file to create a non-executing package file. Access is provided to the non-executing package file where the associated data file is prevented from being executed until data file reputation information is received.

Safe file transmission and reputation lookup

Technologies for an application reputation service to assist users with minimizing their computerized machines' exposure to and infection from malware, including an application reputation service that contains the reputations for elements (e.g., applications) that are known to be non-malicious as well as those known to be malicious. In one example, when a user attempts to install or execute a new application, the service is queried by the user's machine with a set of identities for the element. The service determines the reputation of the application by referencing a knowledge base of known reputations and returns an indication (e.g., an overall rating, or a flag) of how safe that application would be to install and run on the user's computer.

Application reputation service

Technologies described herein relate to estimating engagement of a person with respect to content being presented to the person. A sensor outputs a stream of data relating to the person as the person is consuming the content. At least one feature is extracted from the stream of data, and a level of engagement of the person is estimated based at least in part upon the at least one feature. A computing function is performed based upon the estimated level of engagement of the person.

Estimating engagement of consumers of presented content

The claimed subject matter is directed to the use of an application reputation service to assist users with minimizing their computerized machines' exposure to infection from malware. The claimed subject matter provides an application reputation service that contains the reputations for elements that are known to be non-malicious as well as those known to be malicious. One embodiment is implemented as a method to determine the reputation of an element (e.g., an application). When a user attempts to install or execute a new application, the Application Reputation Service is queried by the user's machine with a set of identities for the element. The reputation of the application is determined by referencing a knowledge base of known reputations and returns an indication (e.g., an overall rating, or a flag) of how safe that application would be to install and run on the user's computer.

Safe file transmission and reputation lookup. As a part of the safe file transmission and reputation lookup methodology, a data file that is to be made available to a data file receiver is accessed and it is determined whether the data file needs to be provided a protective file. The data file is wrapped in a protective file to create a non-executing package file. Access is provided to the non-executing package file where the associated data file is prevented from being executed until data file reputation information is received.

Determining email filtering type based on sender classification. Incoming email is accessed and a sender of the incoming email is identified. The reputation of the sender of the incoming email is determined. An email sender classification is made based on the reputation of the sender. A determination of the type of filtering operations to be performed by an email filter on the email is made based on the classification. The parameters of the type of filtering operations to be performed are determined. The parameters of the type of filtering operations to be performed are provided to the email filter.

Determining email filtering type based on sender classification

Growing community of inventors

Lynnwood, WA, United States of America

Geoff Hulten