Microsoft Technology Licensing, LLC

Meta Platforms, Inc.

Steris Inc.

Jinsong Yu

Galen Clyde Hunt

William Ben Kunz

Luke Kelly

Christopher Wellington Brumme

Michael Robert Hanson

Heidi Young

Seth Robert Atkinson

Emmanouil Koukoumidis

Benoit F Dumoulin

Rajen Subba

Rushin N Shah

Xin Li

Jeremy Gillmor Kahn

Chandrasekhar Iyer

Martin Taillefer

Fuchun Peng

Ahmed Hassan Mohamed

F Soner Terek

Yaron Weinsberg

Larry Jack Israel

Wen-Ke Chen

Gregory Michael Neverov

Alexander P Riemann

John Joseph Duffy

Michael A Centanni

Efstathios Papaefstathiou

Zhengping Zuo

Chung-Chiun Liu

John A Solaro

Shaun David Cox

Robert Scott Dietz

Jason Todd Hunter

John Joseph Richardson

Davide Massarenti

Stanislav A Oks

Andrew E Goodsell

Dmitry Kakurin

Murray Sargent

Lloyd Hilaiel

Gary Eugene Baalman

Rachel Jordan Engel

Dong Zhao

Guowei Shieh

Jihang Li

Maxim Grabarnik

Fehim Soner Terek

John Richardson

In one embodiment, a method includes receiving a user query inputted on a head-mounted device from the head-mounted device, wherein the user query corresponds to multiple dialog-intents, executing multiple tasks corresponding to the multiple dialog-intents, generating a multi-perspective response by a stitching model based on two or more of execution results of the multiple tasks, wherein the stitching model combines the two or more of the execution results based on natural language processing, and wherein the multi-perspective response comprises a natural-language response combining the two or more execution results, and sending instructions to the head-mounted device for presenting the multi-perspective response on the head-mounted device.

Generating multi-perspective responses by assistant systems

In one embodiment, a method includes receiving a user query associated with dialog-intents at a client system, executing tasks corresponding to the dialog-intents, generating a multi-perspective response by a stitching model based on two or more of execution results of the tasks, wherein the multi-perspective response comprises a natural-language response combining the two or more execution results, and presenting the multi-perspective response at the client system.

In one embodiment, a method includes receiving a sharing request to share content generated during a current dialog session from a client system associated with a first user, identifying one or more content objects associated with the sharing request based on a natural-language understanding module, wherein the one or more content objects were previously generated during the current dialog session, determining one or more second users the sharing request is directed to based on a user profile associated with first user, and sending instructions for accessing one or more of the identified content objects to one or more client systems associated with the one or more second users.

Assisting users with efficient information sharing among social connections

In one embodiment, a method includes receiving, from a client system associated with a first user, a first audio input. The method includes generating multiple transcriptions corresponding to the first audio input based on multiple automatic speech recognition (ASR) engines. Each ASR engine is associated with a respective domain out of multiple domains. The method includes determining, for each transcription, a combination of one or more intents and one or more slots to be associated with the transcription. The method includes selecting, by a meta-speech engine, one or more combinations of intents and slots from the multiple combinations to be associated with the first user input. The method includes generating a response to the first audio input based on the selected combinations and sending, to the client system, instructions for presenting the response to the first audio input.

Speech recognition accuracy with natural-language understanding based meta-speech systems for assistant systems

In one embodiment, a method includes receiving a user query from a client system associated with a first user, determining a plurality of dialog-intents, each being associated with a particular agent of a plurality of agents, executing a plurality of tasks corresponding to the user query via the plurality of agents corresponding to the plurality of dialog-intents, receiving a plurality of execution results corresponding to the plurality of tasks from the plurality of agents, respectively, selecting two or more of the plurality of execution results for combination, generating a multi-perspective response based on the selected execution results by a stitching model, wherein the multi-perspective response comprises a natural-language response combining the selected execution results, and sending instructions for presenting the multi-perspective response to the first user to the client system in response to the user query.

In one embodiment, a method includes receiving a first audio input from a user requesting access to information or a service, sending a request for the information or service to an authentication server, where the request includes first authentication information based on the first audio input and further includes a user identifier associated with the user, receiving a second audio input from a client system of the user, where the second audio input includes a first authentication code to be compared to a second authentication code generated by the authentication server, sending second authentication information based on the second audio input and the user identifier to the authentication server for verification, receiving, from the authentication server, an indication of whether the user is successfully authenticated based on the second authentication information, and providing, to the user, access to the information or service when the user is successfully authenticated.

Secure authentication for assistant systems

Automatically generating code used with device drivers for interfacing with hardware. The method includes receiving a machine readable description of a hardware device, including at least one of hardware registers or shared memory structures of the hardware device. The method further includes determining an operating system with which the hardware device is to be used. The method further includes processing the machine readable description on a code generation tool to automatically generate code for a hardware driver for the hardware device specific to the determined operating system.

Software interface for a hardware device

Build-time resolution and type-enforcing of corresponding references in different code that references the same value. In response to detecting a directive within the code itself that a first reference in first code is to be correlated with a second reference in second code, and in response to detection that the types of the references are the same, a code generation tool generates correlation code that is interpretable to a compiler as allowing a value of a type of the first reference of a compiled-form of the first code to be passed as the same value of the same type of the second reference of a compiled-form of the second code. The first code, the second code, and the generated correlation code may then be compiled. If compilation is successful, this means that the first and second references are already properly resolved as referring to the same value and type-enforced.

Build-time resolving and type checking references

Enforcing limitations on hardware drivers. The method includes from a system kernel, assigning I/O resources to the system's root bus. From the root bus, the method further includes assigning a subset of the I/O resources to a device bus. Assigning a subset of the I/O resources to a device bus includes limiting the device bus to only be able to assign I/O resources that are assigned to it by the root bus. From the device bus, the method includes assigning I/O resources to a device through a device interface.

Capability based device driver framework

Caching of an immutable buffer that has its data and address prevented from changing during the lifetime of the immutable buffer. A first computing entity maintains a cache of the immutable buffer and has a strong reference to the immutable buffer. So long as any entity has a strong reference to the immutable buffer, the immutable buffer is guaranteed to continue to exist for the duration of the strong reference. A second computing entity communicates with the first computing entity to obtain a strong reference to the immutable buffer and thereafter read data from the immutable buffer. Upon reading the data from the cache, the second computing entity demotes the strong reference to a weak reference to the immutable buffer. A weak reference to the immutable buffer does not guarantee that the immutable buffer will continue to exist for the duration of the weak reference.

Zero-copy caching

Communicating between an application and a hardware device. A method includes an application writing data to host physical memory using an application view of the memory. The method further includes mapping the data in the physical memory to a hardware driver view, usable by a hardware driver, without needing to copy the data to a different physical storage location. The method further includes mapping the data to a hardware accessible view accessible by a hardware device without needing to copy the data to a different physical storage location.

DMA channels

Input filters correlate to target components. For a given target component, the input filter defines input validation information. The input filter might also define conversions or transformations to be applied to valid input prior to being provided to the target component. At build time, code is accessed that contains the input validation, conversion and transformation and that identifies the associated target component. The information is then used to construct an input filter. At run time, when an input processing component receives an input, the input processing component identifies the target component, accesses the associated input filter, and uses the information contained in the input filter to determine whether the input is valid, and whether and how to convert and transform the value.

Input filters and filter-driven input processing

The environment and use of an immutable buffer. A computing entity acquires data or generates data and populates the data into the buffer, after which the buffer is classified as immutable. The classification protects the data populated within the immutable buffer from changing during the lifetime of the immutable buffer, and also protects the immutable buffer from having its physical address changed during the lifetime of the immutable buffer. As different computing entities consume data from the immutable buffer, they do so through views provided by a view providing entity. The immutable buffer architecture may also be used for streaming data in which each component of the streaming data uses an immutable buffer. Accordingly, different computing entities may view the immutable data differently without having to actually copy the data.

Immutable sharable zero-copy data and streaming

The present invention extends to methods, systems, and computer program products for protected mode scheduling of operations. Protected mode (e.g., user mode) scheduling can facilitate the development of programming frameworks that better reflect the requirements of the workloads through the use of workload-specific execution abstractions. In addition, the ability to define scheduling policies tuned to the characteristics of the hardware resources available and the workload requirements has the potential of better system scaling characteristics. Further, protected mode scheduling decentralizes the scheduling responsibility by moving significant portions of scheduling functionality from supervisor mode (e.g., kernel mode) to an application.

Protected mode scheduling of operations

Architecture for a word processing application that facilitates operating on mathematical symbols, expressions, and/or equations input to a word processing document, and returning results back to the document. User input to the document in the form of math symbols, expressions or equations is transformed into a format for processing by a math engine. The engine returns one or more operations to the user that can be performed on the input, including calculating mathematical solutions, graphing equations and viewing steps to solving math problems. A user interface allows the user choose from the possible operations and to interactively manipulate input and graphs in the word application. The results can be inserted directly into the document and also be graded automatically.

Math calculation in word processors

Growing community of inventors

Bellevue, WA, United States of America

Jinsong Yu