Abstract: Managing rendering of a graphical user interface is provided. A system receives data packets comprising an input audio signal. The system determines an application identifier and query. The system provides the query to the application to cause the application to generate a second query for transmission to a third-party server, and identify responses to the query. The system intercepts the responses, and generates a keyword based on the responses. The system selects a digital component using the keyword, executes a deduplication process, and determines to add the digital component to the responses. The system constructs a display output using a graphical user interface template that integrates the plurality of responses generated by the application with the digital component, and provides the display output to the computing device for rendering.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for classifying videos using neural networks. One of the methods includes obtaining a temporal sequence of video frames, wherein the temporal sequence comprises a respective video frame from a particular video at each of a plurality time steps; for each time step of the plurality of time steps: processing the video frame at the time step using a convolutional neural network to generate features of the video frame; and processing the features of the video frame using an LSTM neural network to generate a set of label scores for the time step and classifying the video as relating to one or more of the topics represented by labels in the set of labels from the label scores for each of the plurality of time steps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for instance segmentation. In one aspect, a system generates: (i) data identifying one or more regions of the image, wherein an object is depicted in each region, (ii) for each region, a predicted type of object that is depicted in the region, and (iii) feature channels comprising a plurality of semantic channels and one or more direction channels. The system generates a region descriptor for each of the one or more regions, and provides the region descriptor for each of the one or more regions to a segmentation neural network that processes a region descriptor for a region to generate a predicted segmentation of the predicted type of object depicted in the region.

Abstract: Methods, systems, and apparatus include computer programs encoded on a computer-readable storage medium, including a method for providing content. A first call for content to be published on a resource associated with a publisher is received from a requesting device. A determination is made when the publisher has authorized additional content item calls to be automatically generated responsive to receipt of the first call. A first content item responsive to the first call is determined. A message including call parameters is generated that define publisher or serving system criteria for the additional content items. The first content item and message responsive to the first call are provided. A second call for content, in conformance with the call parameters, is received after processing the message. A second different content item that is responsive to the second call is provided for publication on the resource along with the first content item.

Abstract: Systems and methods are disclosed herein for coherently modulating and demodulating coherent optical signals using 5-bit constellations. The 5-bit constellations have improved suitability for short reach optical communications systems. A first 5-bit constellation can be formed from four rings. The first and fourth rings can form quadrature phase-shift keyed four-symbol sub-constellations, and the second and third rings can form phase-shift keyed twelve symbol constellations. Symbols from the third and fourth sub-constellations are arranged to form a square. A second 5-bit constellation can include an inner 16 symbol sub-constellation and an sixteen symbol outer sub-constellation. Each of the symbols of the inner sub-constellation are equally spaced from its nearest neighboring symbols along both the quadrature and in-phase axes. Each of the symbols in the outer constellation has an equal euclidean-distance with its nearest neighboring symbols in the outer sub-constellation.

Abstract: A method includes receiving a training example for a listen-attend-spell (LAS) decoder of a two-pass streaming neural network model and determining whether the training example corresponds to a supervised audio-text pair or an unpaired text sequence. When the training example corresponds to an unpaired text sequence, the method also includes determining a cross entropy loss based on a log probability associated with a context vector of the training example. The method also includes updating the LAS decoder and the context vector based on the determined cross entropy loss.

Abstract: A method of performing speech recognition using a two-pass deliberation architecture includes receiving a first-pass hypothesis and an encoded acoustic frame and encoding the first-pass hypothesis at a hypothesis encoder. The first-pass hypothesis is generated by a recurrent neural network (RNN) decoder model for the encoded acoustic frame. The method also includes generating, using a first attention mechanism attending to the encoded acoustic frame, a first context vector, and generating, using a second attention mechanism attending to the encoded first-pass hypothesis, a second context vector.

Abstract: This document describes methods, devices, systems, procedures, and means of maintaining communication and signaling interfaces during network role transitions, such as a secondary node change or a handover. A source secondary base station holds a source donor Central Unit (Source CU) for its own Distributed Unit (source DU) and for the Distributed Unit(s) of the node base station(s) (node DU). The source secondary base station connects to the node base station via a first F1 interface. The node DU triggers handover of the node base station and connected User Equipment (UE) from the source secondary base station to a target secondary base station. The node base station creates Transport Network Layer (TNL) connectivity toward a target donor Central Unit (Target CU) and exchanges application layer configuration data with the Target CU via a wireless backhaul link implemented utilizing the TNL connectivity.

Abstract: On a computing device, an overview mode is provided to present overview windows of all applications currently running on the computing device. When one or more applications are running in a windowed mode, a first overview window is generated for each of the one or more applications running in the windowed mode; when one or more applications are running in a full-screen mode, a second overview window is generated for each of the one or more applications running in the full-screen mode. The one or more first overview windows in the first space can be arranged in one or more rows in a first overview space, and the one or more second overview windows in the second space in a stack in a second overview space. The arranged overview windows may then be displayed in the overview mode of the computing device.

Abstract: The present disclosure provides systems and methods for implementations of motion paths via pure CSS3 and HTML5, working in all major browsers and requiring no JavaScript. For each motion path degree of freedom (e.g., x translation), the system may insert an additional element into the document object model (DOM) to host its animation. In some implementations, the system may apply an optimization process to fit CSS3 keyframes rules that approximate the ideal motion path trajectory to a predetermined tolerance while minimizing the storage footprint. In addition to supporting CSS3 motion paths, this authoring model retains the ability to supply arbitrary standard CSS3 animations to transform channels, which allows users to, e.g., animate the scale and rotation of an element independent of its progress along a motion path.

Abstract: Systems and methods are described that provide measures of uncertainty for map features provided in association with a vehicle map service. A vehicle map service may include a vehicle map model that is configured to provide map data for a geographic area in association with a vehicle map service. The vehicle map model can have an entity schema that includes a first address space configured to represent a plurality of entities associated with the vehicle map model. The vehicle may service may include an uncertainty model that is configured to represent a plurality of uncertainties associated with the plurality of entities. The uncertainty model can have an uncertainty schema that includes a second address space that is separate from the first address space. Uncertainties in the second address space can point to one or more entities in the first address space.

Abstract: Systems, methods, and apparatus include computer programs encoded on a computer-readable storage medium, including a system for ranking videos. Videos are identified that have been presented at client devices. For each video, session start data is identified that specifies a lead video that initiated presentation to a user during a presentation session. For each lead video, presentation times over multiple user sessions are determined, a scaled presentation time is obtained, user sessions for which the lead video initiated presentation of videos are identified, and an aggregate video presentation time attributable to the lead video is determined. For each given video, a presentation score is determined based on a scaled presentation time of the lead video relative to a sum of the aggregate video presentation times for the lead videos. The videos are ranked based on the presentation scores. A user interface is updated to present the ranked videos.

Abstract: A spatially adaptive quantization-aware deblocking filter is used for encoding or decoding video or image frames. The deblocking filter receives a reconstructed frame produced based on dequantized and inverse transformed coefficients of a video frame or an image frame. The reconstructed frame is filtered according to adaptive quantization field data for the video or image frame. The adaptive quantization field data represents weights applied to quantization values used at different areas of the video or image frame. A number of blocking artifacts remaining within the resulting filtered frame is determined. The adaptive quantization field data is then adjusted based on that number of blocking artifacts. The filtered frame is then filtered according to the adjusted adaptive quantization field data. The resulting re-filtered frame is then output to an output source, such as for transmission, display, storage, or further processing.

Abstract: Methods, systems, and media for performing personalized actions on mobile devices associated with a media presentation device are provided.

Abstract: A method includes identifying a middlebox receiving network flow and communicating with one or more backend virtual machines. The method also includes receiving flow statistics corresponding to the network flow of the middlebox and determining whether the flow statistics satisfy an offload rule. The offload rule indicates when to migrate the network flow from the middlebox to an end host. When the flow statistics satisfy the offload rule, the method also includes migrating the network flow from the middlebox to the end host.