Pencil Learning Technologies, Inc.

Deepmind Technologies Limited

Google Inc.

Amogh Sanjay Asgekar

Ayush Agarwal

Xiaohong Gong

Lucian Florin Cionca

Piyush Prahladka

Harish Chandran

Krishnamurthy Dvijotham

Cynthia Johanson

Anton Zhernov

Arvind Jayasundar

Arturo Bajuelos Castillo

Anton Alexandrov

Xueliang Lu

Carsten Miklos Steinebach

Panchapagesan Krishnamurthy

Sanjeev Jagannatha Rao

Zhaohui Fu

Sai Deepak Talasila

The present disclosure provides systems and methods for managed video conferencing. A system can include a media server comprising a processor, a network interface in communication with a plurality of client devices, and a memory storing a cryptographic key shared with an access control server. The system can receive, from a first client device via the network interface, metadata of a video conferencing session and a token, the token provided to the first client device by the access control server responsive to successful registration of the first client device for the video conferencing session. The system can calculate a hash of the metadata with the cryptographic key. The system can compare the calculated hash to the token. The system can, responsive to the calculated hash matching the token, provide at least one media stream of the video conferencing session to the first client device. Such a stream may be provided to the first client device regardless of what other devices are configured to receive the stream or receive other streams within the system.

Observation of parallel video collaboration sessions

The present disclosure provides systems and methods for extraction of nonverbal communication data from video. A system can include a computing device comprising a processor and a camera. The system can retrieve, from a camera of the computing device, a video stream of a user. The system can select a plurality of individual frames of the video stream. For each of the plurality of individual frames of the video stream, the system can extract a plurality of features and identify, from the extracted plurality of features, a pose of the user. The system can classify, via a neural network from the identified poses of the user for the plurality of individual frames of the video stream, the video stream as showing one of a predetermined plurality of states.

Emotion and action tracking and tracking for video conferencing

The present disclosure provides systems and methods of providing multi-user collaborative interfaces. A computing system may receive a video stream. The video stream may have an overlay interface defined along a first plane. The computing system may identify, in at least one frame, a region defined on a second plane on which a user's hand is oriented. The computing system may determine that a set of features associated with the user's hand within the region corresponds to a draw command. The computing system may identify, from the region, a first coordinate for the draw command based on at least one of the set of features. The computing system may translate the first coordinate to a second coordinate defined on the first plane to which to apply the draw command. The computing system may render a visual element at the second coordinate in the overlay interface.

Multi-user collaborative interfaces for streaming video

The present disclosure provides systems and methods for formatting and generating teaching media for different contexts. A system can receive, from a client computing device, media content having one or more media modalities each having presentation attributes. The system can determine initial formatting rules for the media content based on the presentation attributes of the one or more media modalities. The system can identify a destination format for the media content having a formatting requirement for each of the one or more media modalities of the media content. The system can modify the initial formatting rules for the media content based on the formatting requirement for each of the one or more media modalities. The system can generate an information resource having the destination format based on the modified formatting rules for the media content.

Systems and methods for generating educational fluid media

The present disclosure provides systems and methods for generating deeply integrated presentation gadgets. A system can maintain a plurality of application interfaces each having one or more data fields for presenting content. The system can maintain an information resource having data objects. Each of the data objects can have a content attribute. The system can extract each data object from the information resource based on the respective content attribute. The system can generate a presentation data structure having fields for the information resource based on the extracted data objects. The system can select an application interface for presentation with the information resource. The application interface can populate the one or more data fields of the application interface. The system can present the application interface with the information resource at a client device based on the one or more data fields of the application interface.

Systems and methods for deeply integrated presentation gadgets

The present disclosure provides systems and methods for building an inventory database with automatic labeling. A system can maintain a hierarchical concept tree including labels. Each of the labels is associated with a set of attributes and a respective embedding. The system can receive, from a provider device, a request to generate labels for an item of media content. The request can include a request attribute. The system can generate, using a gated categorical model, document embeddings for the item of media content. The system can select a subset of the labels based on the request attribute. The system can determine a respective label score for each label of the subset of the labels based on the document embeddings and the respective embedding of the label. The system can provide a selected label of the subset of the labels based on the respective label score of the selected label.

Systems and methods for building an inventory database with automatic labeling

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a neural network used to generate a ranking score for a network input. One of the methods includes generating training data and training the neural network on the training data. The training data includes a plurality of training pairs. The generating comprising: obtaining data indicating that a plurality of training network inputs were displayed in a user interface according to a presentation order, obtaining data indicating that a first training network input of the plurality of training network inputs has a positive label, determining that a second training network input of the plurality of training network inputs (i) has a negative label and (ii) is higher than the first training network input in the presentation order, and generating a training pair that includes the first training network input and the second training network input.

Pairwise ranking using neural networks

The present disclosure provides systems and methods for encoding and decoding a machine-readable document. A system can include a computing device comprising a processor and a memory. To encode each entry on a document, the processor can receive an identification of an entry, identify an entry format, and generate an entry fingerprint. The processor can calculate a hash value from the identification, the entry format, and the entry fingerprint, which can then be stored at an address corresponding to the hash value. The processor can generate an optical code, and print the entry and the optical code on the document. To decode the document, the processor can extract the hash value, a first entry identifier, and a user identifier from the optical code. The processor can retrieve a second entry identifier and determine a match with the first entry identifier. The processor can retrieve entry coordinates and extract the entry.

Autograder friendly printable worksheets

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for re-ranking a collection of documents according to a first metric and subject to a constraint on a function of one or more second metrics. One of the methods includes: obtaining, for each document in the first collection of documents, a respective first metric value corresponding to the first metric and respective one or more second metric values corresponding to the one or more second metrics; re-ranking the first collection of documents, comprising: determining the constraint on the function of one or more second metrics by computing a first threshold value using a variable threshold function that takes as input second metric values for the documents in the first collection of documents; and determining the re-ranking for the first collection of documents by solving a constrained optimization for the first metric constrained by the first threshold value.

Variable thresholds in constrained optimization

Aspects of the subject technology relate to mixing an aggregated content data structure into a content stream. A first plurality of aggregated content data structures is be received, each comprising a group of content items for display in a content stream of a user and being associated with a respective score. A first aggregated content data structure may be selected from among the first plurality of aggregated content data structures based on the scores and predetermined type target percentages associated with the user. The first aggregated content data structure may be mixed into a content stream associated with a user, and the content stream provided for display.

Growing community of inventors

Palo Alto, CA, United States of America

Amogh Sanjay Asgekar