Microsoft Technology Licensing, LLC

Amazon Technologies, Inc.

Matthew James Wood

Poorna Chand Srinivas Perumalla

Swaminathan Sivasubramanian

Nafea Bshara

Sudipta Sengupta

Leo Parker Dirac

Andrea Olgiati

Malia Morioka Douglas

Zeyad Rajabi

Dominic Rajeev Divakaruni

Bratin Saha

Tali Roth

Robin Winter Troy

Jonathan Bailor

Michael Anthony Affronti

Janet L Schorr

Scott M Stiles

Mark Sunderland

Patrick Michael Conlan

Brian Robert Marshall

Long Gao

Jason Lenox Copeland

Vinayak Ashutosh Agarwal

Oliver Zheng

Robert Ion Gavrila

David De La Brena

Eric Broberg

Naga Krishna Teja Komma

Jodie Draper

Taylor Hopper

Ricardo Elizondo Costa

Jiajun Sun

Rahul Sharma

Tristan Davis

David Robert Thomas

Ryan Edward Gregg

Andrew M Himberger

Aditi Desai

Nagajyothi Nookula

Sarah Michelle Filman

Charles Duze

Jalaja C Kurubarahalli

Pragya Agarwal

Rajankumar Singh

Karan Hemant Kothari

Jodie P B Draper

Amit Wadhwani

Pam Teh

Samuel Oshin

Cory Pruce

Jun Wu

Eftiquar Shaikh

Daniel Evans

Umang Wadhwa

Mark Klunder

Zdravko Pantic

Implementations detailed herein include description of a computer-implemented method. In an implementation, the method at least includes attaching a first set of one or more accelerator slots of an accelerator appliance to an application instance of a multi-tenant provider network according to an application instance configuration, the application instance configuration to define per accelerator slot capabilities to be used by an application of the application instance, wherein the multi-tenant provider network comprises a plurality of computing devices configured to implement a plurality of virtual compute instances, and wherein the first set of one or more accelerator slots is implemented using physical accelerator resources accessible to the application instance; while performing inference using the loaded machine learning model of the application using the first set of one or more accelerator slots on the attached accelerator appliance, managing resources of the accelerator appliance using an accelerator appliance manager of the accelerator appliance.

Fault-tolerant accelerator based inference service

Techniques for machine learning (ML) model training and deployment using model building blocks via graphical user interfaces (GUIs) are described. Users can use a GUI provided by an electronic device to select and configure ML aspects for one or more ML models to be trained using identified training data. The electronic device can send a request to cause a model construction service to train one or more ML models based on the user configuration, return results of the training to the user within the GUI, and deploy one or more of the ML models.

Machine learning model creation via user-configured model building blocks

Techniques for training a machine learning model based on captured images are described. A method described include filtering a first set of collected images using one or more machine learning models; labeling the first set of filtered, collected images using a data labeling service using a service of the provider network; training a machine learning model from a machine learning algorithm using the first set of filtered, collected images using a service of the provider network; and causing deployment of the trained machine learning model onto a device.

Determining data relevance and labeling, model updating, and deployment

Implementations detailed herein include description of a computer-implemented method. In an implementation, the method at least includes receiving an application instance configuration, an application of the application instance to utilize a portion of an attached accelerator during execution of a machine learning model and the application instance configuration including: an indication of the central processing unit (CPU) capability to be used, an arithmetic precision of the machine learning model to be used, an indication of the accelerator capability to be used, a storage location of the application, and an indication of an amount of random access memory to use.

Attached accelerator based inference service

Implementations detailed herein include description of a computer-implemented method. In an implementation, the method at least includes receiving an application instance configuration, an application of the application instance to utilize a portion of an attached accelerator during execution of a machine learning model and the application instance configuration including an arithmetic precision of the machine learning model to be used in determining the portion of the accelerator to provision; provisioning the application instance and the portion of the accelerator attached to the application instance, wherein the application instance is implemented using a physical compute instance in a first location, wherein the portion of the accelerator is implemented using a physical accelerator in the second location; loading the machine learning model onto the portion of the accelerator; and performing inference using the loaded machine learning model of the application using the portion of the accelerator on the attached accelerator.

Attached accelerator selection and placement

Techniques for partitioning data flow operations between execution on a compute instance and an attached accelerator instance are described. A set of operations supported by the accelerator is obtained. A set of operations associated with the data flow is obtained. An operation in the set of operations associated with the data flow is identified based on the set of operations supported by the accelerator. The accelerator executes the first operation.

Framework integration for instance-attachable accelerator

Implementations detailed herein include description of a computer-implemented method. In an implementation, the method at least includes provisioning an application instance and portions of at least one accelerator attached to the application instance to execute a machine learning model of an application of the application instance; loading the machine learning model onto the portions of the at least one accelerator; receiving scoring data in the application; and utilizing each of the portions of the attached at least one accelerator to perform inference on the scoring data in parallel and only using one response from the portions of the accelerator.

Attached accelerator scaling

Systems, methods, and apparatuses for determining data relevance and labeling, model updating, and deployment

Techniques for an electronic montage system are described. An apparatus may comprise a logic device arranged to execute a montage application comprising an authoring component operative to provide a presentation surface having multiple presentation tiles, receive control directives to associate content files with presentation tiles, generate tile objects for the content files based on content file types for the content files, and store the presentation surface and tile objects as a montage. Other embodiments are described and claimed.

Techniques for electronic aggregation of information

Technologies are described for rich visualization of embedded objects, attachments, and links within emails and similar forms of online conversations. In some examples, the embedded objects, attachments, and links may be analyzed and parsed to generate teasers, which may include enhanced images, summary information, and comparable features. The teasers may be presented within a body of an email or in a dedicated area of a conversation user interface and include interactive features such as presentation of detailed information upon selection without opening and processing individual emails.

Interactive preview teasers in communications

Technologies are described for modification of conversation presentation for enhanced user experience. Parts of the conversation may be detected and analyzed for their properties such as who sent a message, who received, when, was anyone left out from or added to the conversation, etc. The extracted properties and analyzed information may then be used to reverse a chronological order of the conversation and present it using interactive features such as collapsing and expanding parts of the presented information, providing a summary, removing unnecessary content, and comparable features.

Conversation modification for enhanced user interaction

Techniques to manage collaborative documents are described. An application program may comprise a document render component to render a first document instance of a collaborative document, a document share component to receive a document update list comprising a set of change records for a second document instance of the collaborative document, and modify properties of one or more constructs for the first document instance based on the change records to form a merged document instance of the collaborative document, and an undo manager component to manage a local undo stack for the first document instance, the local undo stack comprising a set of undo records, and determine whether to preserve one or more of the undo records of the local undo stack after formation of the merged document instance based on one or more of the change records. Other embodiments are described and claimed.

Techniques to manage collaborative documents

Techniques for an electronic montage system are described. An apparatus may comprise a logic device arranged to execute a montage application comprising a presentation component. The presentation component may be operative to detect when a display size or resolution has changed from one used for an original layout of a montage. The presentation component may reposition, or reflow, the tiles in the montage in the changed display in a way that preserves the spatial characteristics of the tiles to each other. An authoring component may reflow tiles automatically when a tile is deleted from a montage to minimize blank space between tiles. Other embodiments are described and claimed.

Techniques for dynamic layout of presentation tiles on a grid

Techniques to determine network storage for sharing media files are described. An apparatus may comprise a processor and a memory communicatively coupled to the processor. The memory may store an application having a media share component operative to receive a request to associate a media file with a document having a document network storage location. The media share component may comprise multiple software modules, including a file manager module. The file manager module may be operative to determine a media file network storage location for the media file based on the document network storage location. Other embodiments are described and claimed.

Techniques to determine network storage for sharing media files

Growing community of inventors

Seattle, WA, United States of America

Matthew James Wood