Amazon Technologies, Inc.

Steve Huynh

James Joseph Jesensky

Isaac W Oates

Vinay P Vaidya

Paul Martin

Michael James Bartholomew

Adam Douglas Morley

Ronald Diaz

Owen Cummings

Karl Gohde

Erik Martin Zigman

Antonio Gabriel Perez De Tejada Martinez

Graham Pedersen

James Christopher Sorenson

Charles Benjamin Waggoner

James Mutton

James Christopher Sorenson, Iii

Benjamin Raphael Forman

Prasanth Saraswatula

Hani Suleiman

Andrew James McVeigh

Systems, devices, and methods are provided for processing video streams. Metadata is extracted from an input video stream and processed using a video stream analyzer. The extracted metadata may be correlated along a time dimension. A metadata graph is generated based on relationships between various information present in the video stream as well as external fact sources. Machine learning models may be trained to receive an input phrase, determine a graph query from the input, and determine an output by traversing the metadata graph according to the graph query.

Machine learning techniques for processing video streams using metadata graph traversal

Systems, devices, and methods are provided for processing images using machine learning. Features may be obtained from an image using a residual network, such as ResNet-. Features may be analyzed using a classification model such as K-nearest neighbors (K-NN). Features and metadata extracted from images may be used to generate other images. Templates may be used to generate various types of images. For example, assets from two images may be combined to create a third image.

Generating images using image assets extracted from other images

Some implementations include methods for generating a portable entitlement for a digital asset and may include generating a portable entitlement to a digital asset based on a request initiated by a first user having an entitlement to the digital asset, the portable entitlement to enable the first user to access the digital asset using a second computing device of a second user, the request initiated using a first computing device of the first user, the portable entitlement having a limited lifetime; and terminating the second computing device from accessing the digital asset based on one or more of determining that a proximity between the first and second computing devices violate the distance threshold and the lifetime of the portable entitlement has expired.

Portable entitlement

Systems and methods are provided herein for synchronizing content across multiple user devices. Content packets may be individually provided to user devices associated with a group. Each content packet may include a portion of content (e.g., media content) and a content index (e.g., an index that indicates a sequence by which the content packets are to be presented). The user device may begin to play the content packets according to the sequence. The user device may receive a particular content index identifying a particular content packet that occurs latest in the sequence and that has been received by all of the devices of the group. The user device may modify playback of the content packets based at least in part on the particular content index to synchronize playback of the content at the user device with respect to playback at the one or more devices.

Techniques for synchronizing content playback across devices

Embodiments of the present disclosure are directed to, among other things, a system that may receive a stream of data content that corresponds with a content classification. A machine learning (ML) model may be selected based at least in part on the metadata and/or content classification. The data content may be inputted to the ML model and an output from the ML model may indicate that data content is associated with the content classification, a start location, and an end location of the stream of data content. The stream of data content may be truncated or cut by removing portions of the stream of data content that fall before the start location and after the end location. The remaining portion of the stream may be stored as a file and provided to the user device.

State detection of live feed

Computer systems of a multimedia service provider may utilize an artificial intelligence model to automatically match a video content with an existing video title of a catalog, allowing the provider to efficiently and accurately process a high volume of video content being received. The artificial intelligence model may be proceed by analyzing video frames from the video content to extract features and then determining the relevance of a set of features to a particular video title of the catalog. Based on the relevance determination, the computer system may associate the video content with the particular video title. In cases where no match with an existing title is determined, the artificial intelligence model may create a new video title based in part on the extracted features.

Matching videos to titles using artificial intelligence

A system capable of identifying and locking resources (e.g., items, tickets, etc.) may map resources to a bit vector and use bit vectors to efficiently identify and lock the resources. For example, tickets to a concert may be mapped to a bit vector and the system may generate an availability bit vector indicating an availability of the seat locations. Using bit vectors, the system may easily identify available seat locations that satisfy a search request and may select from the available seat locations using an optimization technique. For example, the system may select groups of seats that are next to unavailable seat locations or avoid selling groups of seats that would leave only a single seat available. Thus, the system may avoid fragmentation of ticket sales while improving a latency associated with searching for and reserving seat locations.

Method for selling and searching for items

A system capable of identifying and locking resources (e.g., items, tickets, etc.) may map resources to a bit vector and use bit vectors to efficiently identify and lock the resources. For example, tickets to a concert may be mapped to a bit vector and the system may generate an availability bit vector indicating an availability of the seat locations. Thus, the tickets may be reserved by changing a binary value for bits corresponding to individual tickets. In addition, the system may generate bit vectors corresponding to metadata associated with the seat locations (e.g., aisle seat, handicap accessible, first ten rows, back ten rows, etc.), allowing the system to easily identify available seat locations that satisfy the search request. As the bit vectors are relatively small and can be processed easily, the system may improve a latency and reduce an error rate associated with searching for and reserving seat locations.

Techniques for generating and providing phrases are described herein. These techniques may include analyzing one or more sources to generate a first corpus of phrases, each of the phrases for use as an identifier and/or for association with a user. Once a first corpus of phrases has been generated, these phrases may be filtered to define a second corpus of phrases. Phrases of this second corpus may then be suggested to one or more users. In some instances, the phrases suggested to a particular user are personalized to the user based on information previously known about the user or based on information provided by the user.

Generating constructed phrases

Techniques for determining phrases that are related or similar to a source phrase are described herein. These techniques may include analyzing a corpus of text to locate a first set of phrases that differ from a received source phrase by one word. The techniques may then combine these words of the first set of phrases that differ from the source phrase to create a second set of phrases. The techniques may then filter out a portion of the second set of phrases before outputting another portion for use as an identifier, for association with a user account or for any other use.

Determining similar phrases

Techniques for generating and providing phrases are described herein. These techniques may include analyzing one or more sources to generate a statistically improbable phrase, determining words that compose the statistically improbable phrase, inputting the words into an index, and determine phrases associated with the words. The determined phrases may then be presented to a user.

Determining phrases related to other phrases

Techniques for generating and providing phrases are described herein. These techniques may include analyzing one or more sources to generate a first corpus of phrases, each of the phrases for use as an identifier and/or for association with a user for executing a transaction. Once a first corpus of phrases has been generated, these phrases may be filtered to define a second corpus of phrases. Phrases of this second corpus may then be suggested to one or more users. In some instances, the phrases suggested to a particular user are personalized to the user based on information previously known about the user or based on information provided by the user.

Generating and suggesting phrases

Mining phrases for association with a user

Techniques for generating and providing phrases are described herein. These techniques may include analyzing one or more sources to generate a first corpus of phrases, each of the phrases for use as an identifier and/or for association with a user. The first corpus of phrases may be generated based on characteristics associated with phrases. Once a first corpus of phrases has been generated, these phrases may be filtered to define a second corpus of phrases. Phrases of this second corpus may then be suggested to one or more users. In some instances, the phrases suggested to a particular user are personalized to the user based on information previously known about the user or based on information provided by the user.

Growing community of inventors

Seattle, WA, United States of America

Steve Huynh