Abstract: Systems and methods relating to transmitting data between an internetwork of multiple networks. The transmitted data uses a device address that uses a common addressing scheme regardless of network protocols used to implement the multiple networks. The multiple networks includes networks having multiple network protocols, and the multiple networks extend a connection from a home network to a service using a tunnel.

Abstract: Systems and methods are provided for notifying a user regarding movement of a content item which the user has previously expressed interest in, to a new network location. In one or more aspects, a system includes a notification generation component configured to generate a notification indicating a content item has been moved from a first content source to a second content source in response to movement of the content item from the first content source to the second content source. The system further includes a notification posting component configured to send the notification to a device or an account of a user based in part on a subscription of the user to the first content source or the content item.

Abstract: This document describes techniques for fine-motion virtual-reality or augmented-reality control using radar. These techniques enable small motions and displacements to be tracked, even in the millimeter or sub-millimeter scale, for user control actions even when those actions are small, fast, or obscured due to darkness or varying light. Further, these techniques enable fine resolution and real-time control, unlike conventional RF-tracking or optical-tracking techniques.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a sequence generation neural network. One of the methods includes obtaining a batch of training examples; for each of the training examples: processing the training network input in the training example using the neural network to generate an output sequence; for each particular output position in the output sequence: identifying a prefix that includes the system outputs at positions before the particular output position in the output sequence, for each possible system output in the vocabulary, determining a highest quality score that can be assigned to any candidate output sequence that includes the prefix followed by the possible system output, and determining an update to the current values of the network parameters that increases a likelihood that the neural network generates a system output at the position that has a high quality score.

Abstract: Systems and methods for providing a guaranteed batch pool are described, including receiving a job request for execution on the pool of resources; determining an amount of time to be utilized for executing the job request based on available resources from the pool of resources and historical resource usage of the pool of resources; determining a resource allocation from the pool of resources, wherein the resource allocation spreads the job request over the amount of time; determining that the job request is capable of being executed for the amount of time; and executing the job request over the amount of time, according to the resource allocation.

Abstract: Systems and methods for providing a program listing include storing user profile data and a user identifier for a user; storing an association of the user identifier with user identifiers for each of the plurality of social contacts of the user; receiving program identifier data representing programs currently being viewed by the social contacts; ranking the program identifier data for each of the plurality of social contacts based at least in part on the user profile data; and sending display data representing the program identifier data for display in an order based on the ranking.

Abstract: The present invention relates to annotating images. In an embodiment, the present invention enables users to create annotations corresponding to three-dimensional objects while viewing two-dimensional images. In one embodiment, this is achieved by projecting a selecting object onto a three-dimensional model created from a plurality of two-dimensional images. The selecting object is input by a user while viewing a first image corresponding to a portion of the three-dimensional model. A location corresponding to the projection on the three-dimensional model is determined, and content entered by the user while viewing the first image is associated with the location. The content is stored together with the location information to form an annotation. The annotation can be retrieved and displayed together with other images corresponding to the location.

Abstract: A computer-implemented method for executing a software application in a virtual machine executing on a data processing device includes receiving software code for a software application, determining application programming interfaces referenced by the software code, determining portions of an operating system accessed by the software code and creating an application container in the virtual machine. The method also includes application programming interfaces referenced by the software code inside the application container, portions of the operating system accessed by the software code inside the application container and executing the software application inside the application container on the virtual machine.

Abstract: Methods, systems, and apparatus include computer programs encoded on a computer-readable storage medium, including a method for determining information. A first user device for presenting content to a user is identified. A second co-located device is identified. A determination is made, based on one or more criteria, to use audible or inaudible sound emitted from the first device to ascertain whether audio controls on the first device and a speaker associated with the first device are enabled. A token that is to be played on the first device is identified. A notification is provided to the second co-located device to listen for the token emitted by the first device. The token for playing on the first device is provided. Information related to a recording is received from the second co-located device. A determination is made whether the first device's audio controls and speakers are enabled.

Abstract: Arrangements for controlling a climate control system are provided. A hazard detector of a group of smart devices may detect a carbon monoxide (CO) alarm condition at the hazard detector. The hazard detector may transmit, via a relatively low-power mesh communication network, an indication of the CO alarm condition to one or more other smart devices. A spokesman node of the relatively low-power wireless communication network may translate the indication of the CO alarm condition from a first wireless communication protocol to a second wireless communication protocol. The spokesman node may transmit the indication of the CO alarm condition to a system controller via a relatively high-power wireless communication network and the second wireless communication protocol. A system controller of the climate control system may transmit a signal to turn off at least part of the climate control system.

Abstract: The present disclosure provides a system for intelligently sampling information, such as location, activities, etc. on device. Sampling and uploading of background context is optimized using machine learning, such that battery usage is reduced, and quality of metrics based on the reported information is maintained or improved. A policy is generated based on the machine learning, the policy dictating how scanning and upload rates should change in response to conditions on the device.

Abstract: Various arrangements are presented for monitoring a resident of a residence. A confidence assessment may be performed based on a plurality of smart home devices in the residence. The residence may be identified as eligible for monitoring of the resident based on the confidence assessment. A learning process may be performed to create an ordinary behavior model. Data that is received from the plurality of smart home devices may be monitored to identify data indicative of behavior considered unusual based on the ordinary behavior model.

Abstract: A method for detecting a hotword includes receiving a sequence of input frames that characterize streaming audio captured by a user device and generating a probability score indicating a presence of a hotword in the streaming audio using a memorized neural network. The network includes sequentially-stacked single value decomposition filter (SVDF) layers and each SVDF layer includes at least one neuron. Each neuron includes a respective memory component, a first stage configured to perform filtering on audio features of each input frame individually and output to the memory component, and a second stage configured to perform filtering on all the filtered audio features residing in the respective memory component. The method also includes determining whether the probability score satisfies a hotword detection threshold and initiating a wake-up process on the user device for processing additional terms.

Abstract: Systems and methods for controlling the display of content of information resources are described herein. The system can receive a request for an information resource from a client device. The system can transmit the information resource that can include a first content portion, a second content portion, and a content display management script. The content display management script can include instructions that can cause the client device to: (a) set a scroll range of the information resource from a first location to a second location in the information resource; (b) display, within a portion of the information resource between the first location and the second location, the first content portion and an ad that can include actionable item; and (c) set, responsive to detecting an interaction on the actionable item, the scroll range from the first location to a third location of the information resource.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating representation of acoustic sequences. One of the methods includes: receiving an acoustic sequence, the acoustic sequence comprising a respective acoustic feature representation at each of a plurality of time steps; processing the acoustic feature representation at an initial time step using an acoustic modeling neural network; for each subsequent time step of the plurality of time steps: receiving an output generated by the acoustic modeling neural network for a preceding time step, generating a modified input from the output generated by the acoustic modeling neural network for the preceding time step and the acoustic representation for the time step, and processing the modified input using the acoustic modeling neural network to generate an output for the time step; and generating a phoneme representation for the utterance from the outputs for each of the time steps.

Abstract: Methods, systems, and media for color palette extraction for videos are provided. In some embodiments, the method comprises: identifying, at a server, a frame of a video content item; clustering pixels of the frame of the video content item based on a color of each of the pixels into a group of clusters; for each of a plurality of clusters in the group of clusters, determining an average color for the cluster; selecting a cluster of the plurality of clusters based on the average color of the cluster; determining a color palette corresponding to the frame of the video content item for one or more user interface elements in which the video content item is to be presented based on the average color of the selected cluster; and transmitting information indicating the color palette to a user device in response to a request to present the video content item.

Abstract: Systems and methods described herein may relate to training an artificial neural network (ANN) using a differentiable Jaccard Loss approximation. An example embodiment may involve obtaining a training image and a corresponding ground truth mask that represents a desired segmentation of the training image. The embodiment may further involve applying an ANN on the training image to generate an output segmentation of the training image that depends on a plurality of weights of the ANN and determining a differentiable Jaccard Loss approximation based on the output segmentation of the training image and the ground truth mask. The embodiment also involves modifying one or more weights of the ANN based on the differentiable Jaccard Loss approximation and providing a representation of the ANN as modified to a mobile computing device.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for prioritizing network traffic. In one aspect, a method includes collecting, for local networks, traffic data representing communication traffic over the local network for a period of time; determining, for each of the network devices included in the networks, a device type of the network device and at least one device priority value for the network device; training a device prioritization model, using the traffic data and the device priority values of the devices, receiving, by the device prioritization model, for a local network, a list of network devices that are included in the local network; and generating, by the device prioritization model, a prioritization scheme for the local network that prioritizes device traffic among the network devices based on the device types and device priority values of the network devices.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for a circuit configured to perform computations using multiple inputs. The circuit includes multiple adder circuits and a selection circuit that includes multiple input selector. Each adder circuit performs an addition operation using sets of inputs derived from the multiple inputs. The input selectors are configured to select one or more inputs from a set of inputs derived from the multiple inputs based on a sign bit for an input in the set and pass the selected inputs to an adder circuit that generates a sum using the selected inputs. The circuit determines a routing of the sum to another adder circuit based in part on a sign bit for the input in the set of inputs.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for receiving a request from a client to process a computational graph; obtaining data representing the computational graph, the computational graph comprising a plurality of nodes and directed edges, wherein each node represents a respective operation, wherein each directed edge connects a respective first node to a respective second node that represents an operation that receives, as input, an output of an operation represented by the respective first node; identifying a plurality of available devices for performing the requested operation; partitioning the computational graph into a plurality of subgraphs, each subgraph comprising one or more nodes in the computational graph; and assigning, for each subgraph, the operations represented by the one or more nodes in the subgraph to a respective available device in the plurality of available devices for operation.