Abstract: A method includes receiving sound by a first audio unit installed in an electrical outlet, routing audio data corresponding to the received sound from the first audio unit to a second audio unit installed in a second electrical outlet, and sending the audio data to a mobile device using a wireless link between the mobile device and the second audio unit. Routing the audio data may include receiving the audio data from the first audio unit by a third audio unit and routing the audio data to the second audio unit by the third audio unit serving as a router. The data may be routed using table driven routing, on-demand routing or some other appropriate routing protocol. The method may also include performing voice recognition on the audio data and detecting a command word and routing command word data to the second audio unit.

Abstract: Implementations set forth herein relate to an automated assistant that allows third party applications to inject dependencies to leverage automated assistant functions. Furthermore, enabling such dependency injections can allow third party applications to preserve privacy of any application content that is used during execution of automated assistant functions. In some implementations, a third party application can initialize a function with an assistant dependency using parameters that are tagged as private. Initializing a function in such as a way can allow private content communicated between the third party application and the automated assistant to be abstracted for security purposes. The abstracted content can thereafter be communicated to a remote server—such as a server hosting an extensively trained machine learning model. Intelligent output provided by the server can then be incorporated into one or more processes of the third party application without comprising security.

Abstract: Implementing methods to provide a shortened textual summary that includes status information that is most pertinent to the user for a plurality of connected smart devices. The method includes determining a list of current statuses for a plurality of enabled smart devices and filtering the list to remove statuses that may not be of interest to the user. The filtering of the list is based on a current context of the requesting user and one or more previous contexts of the user. The resulting filtered statuses are then converted to textual snippets, summarized, and provided to the user via one or more output devices.

Abstract: A method including determining a gaze direction of a user of a wearable device, capturing an image using a forward-looking camera of the wearable device, detecting a surroundings of the user based on the image, determining whether or not the user is distracted based on the gaze direction and the surroundings, and in response to determining the user is distracted, causing an operation to be performed on the wearable device, the operation configured to cause the user to change the user's attention.

Abstract: Systems and methods of the present disclosure can include a computer-implemented method for efficient machine-learned model training. The method can include obtaining a plurality of training samples for a machine-learned model. The method can include, for one or more first training iterations, training, based at least in part on a first regularization magnitude configured to control a relative effect of one or more regularization techniques, the machine-learned model using one or more respective first training samples of the plurality of training samples. The method can include, for one or more second training iterations, training, based at least in part on a second regularization magnitude greater than the first regularization magnitude, the machine-learned model using one or more respective second training samples of the plurality of training samples.

Abstract: A computerized method for associating cyberthreat actor groups responsible for different cyberthreats is described. The method involves generating a similarity matrix based on content from received clusters of cybersecurity information. Each received cluster of cybersecurity information is assumed to be associated with a cyberthreat. The similarity matrix is composed via an optimized equation combining separate similarity metrics, where each similarity metric of the plurality of similarity metrics represents a level of correlation between at least two clusters of cybersecurity information, with respect to a particular aspect of operations described in the clusters. The method further involves that, in response to queries directed to the similarity matrix, generating a listing of a subset of the clusters of cybersecurity information having a greater likelihood of being associated with cyberthreats caused by the same cyberthreat actor group.

Abstract: The present disclosure selects third party content based on feedback. A selector identifies several content items including first and second content items (or more) responsive to a request. A machine learning engine determines a first feature of the first content item, a second feature of the second content item, and a third feature of the web page or a device associated with the request. The machine learning engine determines, responsive to the first feature and the third feature, a first score for the first content item based on a machine learning model generated using historical signals received from devices via a metadata channel formed from an electronic feedback interface. The machine learning engine determines a second score for the second content item responsive to the second feature and the third feature. A bidding module determines a price for the first content item based on the first and second scores.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for analyzing images for generating query responses. One of the methods includes determining, using a textual query, an image category for images responsive to the textual query, and an output type that identifies a type of requested content; selecting, using data that associates a plurality of images with a corresponding category, a subset of the images that each belong to the image category, each image in the plurality of images belonging to one of the two or more categories; analyzing, using the textual query, data for the images in the subset of the images to determine images responsive to the textual query; determining a response to the textual query using the images responsive to the textual query; and providing, using the output type, the response to the textual query for presentation.

Abstract: Training and/or using both a high-level policy model and a low-level policy model for mobile robot navigation. High-level output generated using the high-level policy model at each iteration indicates a corresponding high-level action for robot movement in navigating to the navigation target. The low-level output generated at each iteration is based on the determined corresponding high-level action for that iteration, and is based on observation(s) for that iteration. The low-level policy model is trained to generate low-level output that defines low-level action(s) that define robot movement more granularly than the high-level action—and to generate low-level action(s) that avoid obstacles and/or that are efficient (e.g., distance and/or time efficiency).

Abstract: Images of a plurality of users are captured concurrently with the plurality of users evincing a plurality of expressions. The images are captured using one or more eye tracking sensors implemented in one or more head mounted devices (HMDs) worn by the plurality of first users. A machine learnt algorithm is trained to infer labels indicative of expressions of the users in the images. A live image of a user is captured using an eye tracking sensor implemented in an HMD worn by the user. A label of an expression evinced by the user in the live image is inferred using the machine learnt algorithm that has been trained to predict labels indicative of expressions. The images of the users and the live image can be personalized by combining the images with personalization images of the users evincing a subset of the expressions.

Abstract: Some implementations process, using warm word model(s), a stream of audio data to determine a portion of the audio data that corresponds to particular word(s) and/or phrase(s) (e.g., a warm word) associated with an assistant command, process, using an automatic speech recognition (ASR) model, a preamble portion of the audio data (e.g., that precedes the warm word) and/or a postamble portion of the audio data (e.g., that follows the warm word) to generate ASR output, and determine, based on processing the ASR output, whether a user intended the assistant command to be performed. Additional or alternative implementations can process the stream of audio data using a speaker identification (SID) model to determine whether the audio data is sufficient to identify the user that provided a spoken utterance captured in the stream of audio data, and determine if that user is authorized to cause performance of the assistant command.

Abstract: Delay reporting information descriptive of an existing delay of the participant computing device is received from each participant computing device of a participant cohort. The existing delay includes a period of time between transmission of audio data from the teleconference computing system and receipt of the audio data by the participant computing device. The participant cohort includes multiple participant computing devices that are co-located within a same area and are connected to a teleconference hosted by the teleconference computing system. A target delay is determined for the participant cohort. The target delay includes a period of time greater than the existing delay of participant computing devices. The target delay is configured to cause synchronized audio playback when implemented by the devices. Target delay information indicative of the target delay is provided to the participant cohort.

Abstract: Techniques are described herein for dialog-based enrollment of individual users for single- and/or multi-modal recognition by an automated assistant, as well as determining how to respond to a particular user's request based on the particular user being enrolled and/or recognized. Rather than requiring operation of a graphical user interface for individual enrollment, dialog-based enrollment enables users to enroll themselves (or others) by way of a human-to-computer dialog with the automated assistant.

Abstract: Recommending an automated assistant action for inclusion in an existing automated assistant routine of a user, where the existing automated assistant routine includes a plurality of preexisting automated assistant actions. If the user confirms the recommendation through affirmative user interface input, the automated assistant action can be automatically added to the existing automated assistant routine. Thereafter, when the automated assistant routine is initialized, the preexisting automated assistant actions of the routine will be performed, as well as the automated assistant action that was automatically added to the routine in response to affirmative user interface input received in response to the recommendation.

Abstract: A method of assigning categories to electronic messages is performed at a computing device having one or more processors and memory. The method includes receiving an electronic message and extracting one or more syntactic pairs of words from content of the electronic message. At least one of syntactic pairs includes a verb in a first sentence and an object in the same sentence, where the verb operates on the object. The method forms a feature vector based on the extracted syntactic pairs and uses the feature vector as input to a classifier. The classifier is trained on a set of training feature vectors, and each training feature vector is associated with a respective category. The method assigns a first category to the electronic message based on output of the classifier and displays the message in a group with other messages associated with the first category.

Abstract: A method for avoiding undesired message transmissions from a user device to a RAN is implemented in a user device. The method includes receiving (202 or 302), from the RAN, a first value of the barring factor, and determining (206 and 210 or 306 and 310), based on a first probability corresponding to the first value, that access barring applies to a message transmission. The method also includes suspending (214 or 314) the message transmission while running a first timer (216 or 316) and, after the first timer expires or stops, either (1) if the first value does not correspond to a maximum probability of the different probabilities, determining (222 or 322) that the access barring is alleviated, or (2) if the first value corresponds to the maximum probability, continuing to suspend (224 or 326) the message transmission.

Abstract: A route from a current location of a portable device to a destination is determined, where the route includes a sequence of directed sections. Navigation instructions to guide a user of the portable device along the route to the destination are generated. To this end, candidate navigation landmarks perceptible within a 360-degree range of the current location of the portable device are identified, a navigation landmark disposed in a direction substantially opposite to the direction of the first one in the sequence of directed sections is selected from among the candidate navigation landmarks, and an initial instruction in the navigation instructions is generated and provided via a user interface of the portable device. The initial instruction references the selected navigation landmark.

Abstract: Methods, systems, and apparatus, for associating actions with events that are determined independent of the action type of the action. A system receives, from a user device, identification of an action to be taken, the action being one of a plurality of action types, and an identification of a first event that is determined independent of the action type and for which the occurrence of the first event may be determined by the system. They system associates the action and the first event with instructions for determining if the first event has occurred, and then determines whether the first event has occurred. In response to determining that the first event has occurred, the system causes the action to occur.