Abstract: An electronic device can include a first image sensor configured to capture a first image of a field of view and a second image sensor configured to capture a second image of the field of view. The electronic device can include a color filter adjacent to the second image sensor such that the field of view is viewable by the second image sensor through the color filter. The first image can have a first pixel resolution. The second image can have a second pixel resolution. The electronic device can include a controller configured to determine a third image based on luminance content of the first image and color content of the second image. The third image can have a third pixel resolution indicative of a spatial resolution of the first image and a spectral resolution of the second image.

Abstract: Methods, systems, and media for presenting media content items belonging to a media content group are provided.

Abstract: Methods and systems for learning long-term dependencies in recurrent neural networks. In one aspect, a neural network system is configured to receive a respective input for each of a plurality of time steps and to generate a respective output for each time step, the neural network system comprising one or more recurrent neural network layers, wherein, for each of the time steps, each of the recurrent neural network layers is configured to receive a layer input for the time step; apply an input weight matrix to the layer input to generate a first output; apply a recurrent weight matrix to a hidden state of the recurrent neural network layer for the time step to generate a second output; combine the first and second outputs to generate a combined output; and apply a rectified linear unit activation function to the combined output to generate a layer output for the time step.

Abstract: A computer-implemented method can include identifying a first set of text samples that include a particular potentially offensive term. Labels can be obtained for the first set of text samples that indicate whether the particular potentially offensive term is used in an offensive manner. A classifier can be trained based at least on the first set of text samples and the labels, the classifier being configured to use one or more signals associated with a text sample to generate a label that indicates whether a potentially offensive term in the text sample is used in an offensive manner in the text sample. The method can further include providing, to the classifier, a first text sample that includes the particular potentially offensive term, and in response, obtaining, from the classifier, a label that indicates whether the particular potentially offensive term is used in an offensive manner in the first text sample.

Abstract: Systems and methods for determining vehicle crowdedness are provided. A method can include obtaining, by a computing system, real-time location data including information identifying a real-time location corresponding to a transit station for each of a plurality of user computing devices; determining, by the computing system using a vehicle crowdedness model, a vehicle crowdedness for one or more transit vehicles departing from one or more transit stations based at least in part on the real-time location data; and communicating, by the computing system, data indicative of the vehicle crowdedness for the one or more transit vehicles to a particular user computing device. The data indicative of the vehicle crowdedness can include information for displaying the vehicle crowdedness for the one or more transit vehicles on the particular user computing device.

Abstract: Systems and methods are provided for including information in an invitation from a first to connect with a second user that identifies a context regarding how the first user found the second user. A system can include a request component configured to receive, at a first network source, information identifying a second user that a first user has selected to send an invitation to connect with at the first network source, a context component configured to determine a context regarding selection of the second user by the first user, including where the first user found the information identifying the second user, and a generation component configured to generate an invitation that invites the second user to connect with the first user at the first network source. The invitation comprises first user identification information and context information identifying the context regarding the selection of the second user by the first user.

Abstract: Methods, systems, and apparatus for a matrix multiply unit implemented as a systolic array of cells are disclosed. The matrix multiply unit may include cells arranged in columns of the systolic array. Two chains of weight shift registers per column of the systolic array are in the matrix multiply unit. Each weight shift register is connected to only one chain and each cell is connected to only one weight shift register. A weight matrix register per cell is configured to store a weight input received from a weight shift register. A multiply unit is coupled to the weight matrix register and configured to multiply the weight input of the weight matrix register with a vector data input in order to obtain a multiplication result.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content items. In one aspect, a method includes receiving user interaction data specifying user interactions with a publisher resource. A heat map specifying levels of user interaction with various portions of the publisher resource is created. Content item slot performance information specifying one or more performance measures for content items presented in various candidate content item slot locations are identified. One of the candidate content item slot locations is selected based on the heat map and the one or more performance measures. Data that cause presentation of suggestion information that identify the one candidate content item slot location as a suggested content item slot location are generated and output.

Abstract: A method includes obtaining audio data for a long-form utterance and segmenting the audio data for the long-form utterance into a plurality of overlapping segments. The method also includes, for each overlapping segment of the plurality of overlapping segments: providing features indicative of acoustic characteristics of the long-form utterance represented by the corresponding overlapping segment as input to an encoder neural network; processing an output of the encoder neural network using an attender neural network to generate a context vector; and generating word elements using the context vector and a decoder neural network. The method also includes generating a transcription for the long-form utterance by merging the word elements from the plurality of overlapping segments and providing the transcription as an output of the automated speech recognition system.

Abstract: Hardware transactions or other techniques, such as custom PCIe handling devices, are used to atomically move pages from one host's memory to another host's memory. The hosts are connected by one or two non-transparent bridges (NTBs), which make each host's memory and devices available to the other, while allowing each host to reboot independently.

Abstract: A method includes accessing, by a server computer, a plurality of electronic mail messages. The method also includes identifying, by the server computer, at least one of the electronic mail messages as a promoted electronic mail message. The method further includes outputting, for display at a client computing device, an electronic mail interface wherein the promoted electronic mail message is represented in a first position of a display order and a remainder of the electronic mail messages are represented subsequent to the promoted electronic mail message in the display order.

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: Techniques and apparatuses are described that implement a smartphone-based radar system capable of detecting user gestures using coherent multi-look radar processing. Different approaches use a multi-look interferometer or a multi-look beamformer to coherently average multiple looks of a distributed target across two or more receive channels according to a window that spans one or more dimensions in time, range, or Doppler frequency. By coherently averaging the multiple looks, a radar system generates radar data with higher gain and less noise. This enables the radar system to achieve higher accuracies and be implemented within a variety of different devices. With these accuracies, the radar system can support a variety of different applications, including gesture recognition or presence detection.

Abstract: This document describes techniques and systems that enable a smartphone-based radar system for determining user intention in a lower-power mode. The techniques and systems use a radar field to enable the smartphone to accurately determine the presence or absence of a user and further determine the intention of the user to interact with the smartphone. Using these techniques, the smartphone can account for the user's nonverbal communication cues to determine and maintain an awareness of users in its environment, and only respond to direct interactions once a user has demonstrated an intention to interact, which preserves battery power. The smartphone may determine the user's intention by recognizing various cues from the user, such as a change in position relative to the smartphone, a change in posture, or by an explicit action, such as a gesture.

Abstract: The present disclosure is directed to systems and methods of configuring a resource for network traffic analysis. An agent executed by a computing device receives an indication to record a browsing session. The agent records network activity data of the browsing session. The agent generates a data file with the recording. The agent transmits the data file and a request to generate a report to an analytics server. The analytics server receives the request and extracts the network activity data from the data file. The analytics server applies an event processing protocol to the network activity data. The analytics server generates a report and transmits the report to the client device for display.

Abstract: A method for training hotword detection includes receiving a training input audio sequence including a sequence of input frames that define a hotword that initiates a wake-up process on a device. The method also includes feeding the training input audio sequence into an encoder and a decoder of a memorized neural network. Each of the encoder and the decoder of the memorized neural network include sequentially-stacked single value decomposition filter (SVDF) layers. The method further includes generating a logit at each of the encoder and the decoder based on the training input audio sequence. For each of the encoder and the decoder, the method includes smoothing each respective logit generated from the training input audio sequence, determining a max pooling loss from a probability distribution based on each respective logit, and optimizing the encoder and the decoder based on all max pooling losses associated with the training input audio sequence.

Abstract: This specification discloses computer-based systems, methods, devices, and other techniques for estimating the pose of a device, including estimating the pose based on images captured by a set of image sensors disposed around the device's periphery. Some implementations include a system that obtains visual data representing at least one image captured by one or more image sensors of a mobile device. The at least one image show an environment of the mobile device, and the one or more image sensors are located at respective corners of the mobile device, or at other locations around its periphery. The system processes the visual data to determine a pose of the mobile device. Further, the system can determine a location of the mobile device in the environment based on the pose, and can present an indication of the location of the mobile device in the environment.