Abstract: Methods and systems for receiving a request to implement a neural network comprising an average pooling layer on a hardware circuit, and in response, generating instructions that when executed by the hardware circuit, cause the hardware circuit to, during processing of a network input by the neural network, generate a layer output tensor that is equivalent to an output of the average pooling neural network layer by performing a convolution of an input tensor to the average pooling neural network layer and a kernel with a size equal to a window of the average pooling neural network layer and composed of elements that are each an identity matrix to generate a first tensor, and performing operations to cause each element of the first tensor to be divided by a number of elements in the window of the average pooling neural network layer to generate an initial output tensor.

Abstract: A system enables universal remote media control across multiple platforms, devices, and users. A protocol allows a media controller (such as computer or mobile device) to have access to running activities (such as a television show playing on a media player). The protocol does not require installation of a specific platform application, or a specific user-login/pairing of the media controller with the media player. The system receives commands from a media controller device using a common interface, provides the commands to a remote media player, loads new media content into the media player, based on the commands, and receives state notifications from the media player. The commands and the state notifications include a namespace and a message payload represented as a two-element script array. Another user or device can control the media player using the common interface, for example provided by a browser application on the media controller device.

Abstract: A method for converting alternating current (AC) power to direct current (DC) power in a non-isolated power converter includes receiving a three-phase power supply, transforming the three phase power supply into six voltage phases, half-wave rectifying the AC current, applying a power factor correction to achieve DC power, and outputting a DC power signal. The three-phase power supply has an AC current. The six voltage phase is transformed at a secondary side of a three-phase distribution transformer, which includes a center tap located at the secondary side of the three-phase distribution transformer and one or more AC wire conductors. The AC wire conductors carry the transformed power supply. The half-wave rectification occurs at the secondary side of the three-phase distribution transformer. An arrangement of rectifier diodes on the AC wire conductors accomplishes the half-wave rectification. The output DC power signal has an output voltage at a DC output.

Abstract: A system mounted within eyewear or headwear to unobtrusively produce and track reference locations on the surface of one or both eyes of an observer is provided to improve the accuracy of gaze tracking. The system utilizes multiple illumination sources and/or multiple cameras to generate and observe glints from multiple directions. The use of multiple illumination sources and cameras can compensate for the complex, three-dimensional geometry of the head and the significant anatomical variations of the head and eye region that occurs among individuals. The system continuously tracks the initial placement and any slippage of eyewear or headwear. In addition, the use of multiple illumination sources and cameras can maintain high-precision, dynamic eye tracking as an eye moves through its full physiological range.

Abstract: A computerized method for automatically generating display content includes receiving a uniform resource locator, wherein the uniform resource locator specifies a landing resource and extracting visual information from the landing resource, wherein the visual information defines one or more images, texts, and colors displayed on the landing resource. The method further includes selecting one or more images, one or more text snippets, and one or more colors based on the visual information extracted from the landing resource, generating a layout for a content item based on one or more of the selected images or selected text snippets, and assembling the content item by applying the selected images, the selected text snippets, and the selected colors to the generated layout.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating signed addresses. One of the methods includes receiving, by a component from a device, a plurality of first requests, each first request for a physical address and including a virtual address, determining, by the component, a first physical address using the virtual address, generating a first signature for the first physical address, and providing, to the device, a response that includes the first signature, receiving, from the device, a plurality of second requests, each second request for access to a second physical address and including a second signature, determining, by the component for each of the plurality of second requests, whether the second physical address is valid using the second signature, and for each second request for which the second physical address is determined to be valid, servicing the corresponding second request.

Abstract: Multiple image verification challenges can be used to identify the location of an object within an initial image. For instance, a first set of tiles is generated using the initial image. This first set is provided to a client computing device for display in a first verification challenge requesting that the user select tiles including the object. In response, a user selection of tiles of the first set is received. These selected tiles are then used to generate a second set of tiles corresponding to a sub-portion of the initial image. The second set of tiles is provided to a client computing device for display in a second verification challenge. In response to the second verification challenge, a user selection of tiles of the second set is received. This user selection of tiles of the second set is then used to determine a location of the object in the image.

Abstract: Example systems and methods are disclosed for determining the direction of an actor based on image data and sensors in an environment. The method may include receiving point cloud data for an actor at a location within the environment. The method may also include receiving image data of the location. The received image data corresponds to the point cloud data received from the same location. The method may also include identifying a part of the received image data that is representative of the face of the actor. The method may further include determining a direction of the face of the actor based on the identified part of the received image data. The method may further include determining a direction of the actor based on the direction of the face of the actor. The method may also include providing information indicating the determined direction of the actor.

Abstract: A computer-implemented user notification method includes displaying, in a status area near a perimeter of a graphical interface, a notification of a recent alert event for a mobile device, receiving a user selection in the status area, and in response to the receipt of the user selection, displaying, in a central zone of the graphical interface, detail regarding a plurality of recent messaging events for the mobile device.

Abstract: A method for modifying access rights to electronically stored files linked in a draft electronic communication stored at a client device includes receiving, at a server from the client device, information relating to a plurality of recipients of the draft electronic communication, information relating to a sender of the draft electronic communication, and a first link, where the first link represents a first file. The method further includes determining, at the server, for the sender and for each recipient in the plurality of recipients, the access rights to the first file. The method further includes generating, at the server, a plurality of options for the sender to modify the access rights for the plurality of recipients, where the plurality of options is based on the access rights to the first file, and sending the plurality of options from the server to the client device.

Abstract: Example systems and methods are disclosed for associating detected attributes with an actor. An example method may include receiving point cloud data for a first actor at a first location within the environment. The method may include associating sensor data from an additional sensor with the first actor based on the sensor data being representative of the first location. The method may include identifying one or more attributes of the first actor based on the sensor data. The method may include subsequently receiving a second point cloud representative of a second actor at a second location within the environment. The method may include determining, based on additional sensor data from the additional sensor, that the second actor has the one or more attributes. The method may include providing a signal indicating that the first actor is the second actor based on the second actor having the one or more attributes.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for receiving a sequence representing an utterance, the sequence comprising a plurality of audio frames; determining one or more warping factors for each audio frame in the sequence using a warping neural network; applying, for each audio frame, the one or more warping factors for the audio frame to the audio frame to generate a respective modified audio frame, wherein the applying comprises using at least one of the warping factors to scale a respective frequency of the audio frame to a new respective frequency in the respective modified audio frame; and decoding the modified audio frames using a decoding neural network, wherein the decoding neural network is configured to output a word sequence that is a transcription of the utterance.

Abstract: Methods, systems, and apparatus, including computer program products, for establishing preferences for sites that are explicitly identified by a user by either a textual input or an acceptance in response to a recommendation, and adjusting the presentation of search results based on these preferences.

Abstract: Devices and methods are provided for eye-tracking, e.g., including a freeform optical assembly and/or a modular design. In an exemplary embodiment, a device and method are provided that includes a wearable device on a user's head, the wearable device including a scene camera oriented to capture images of the user's surroundings. The user may perform a predetermined action with the user's eye to activate a photo feature of the wearable device, gaze at a region within the user's surroundings, the wearable device determining a focal point and limited field-of-view for the camera imaging field based on the center point, and activate the camera to capture an image of the limited field-of-view centered around the focal point.

Abstract: A computer-implemented method of multisensory speech detection is disclosed. The method comprises determining an orientation of a mobile device and determining an operating mode of the mobile device based on the orientation of the mobile device. The method further includes identifying speech detection parameters that specify when speech detection begins or ends based on the determined operating mode and detecting speech from a user of the mobile device based on the speech detection parameters.

Abstract: Apparatus and methods are provided for application layer optimization in a modern data network. The optimization incorporates variable rate transmission across one or more optical data channels. Data throughput is maximized by enabling quality of service profiles on a per transmission channel basis. According to one aspect, a system is provided in which the application layer is aware of and controls the underlying transmission rate and quality of the transmission. This enables the system to fully utilize the transmission capacity of the channel. The application layer may map different applications to different transmission classes of service. The services can be classified based on data throughput rate, guaranteed error rates, latency and cost, among other criteria. This provides flexibility to the application layer to map some loss tolerant applications to a lower cost (per bit) transmission class.