Abstract: Methods and systems for determining an individual gaze value are disclosed herein. An exemplary method involves: (a) receiving gaze data for a first wearable computing device, wherein the gaze data is indicative of a wearer-view associated with the first wearable computing device, and wherein the first wearable computing device is associated with a first user-account; (b) analyzing the gaze data from the first wearable computing device to detect one or more occurrences of one or more advertisement spaces in the gaze data; (c) based at least in part on the one or more detected advertisement-space occurrences, determining an individual gaze value for the first user-account; and (d) sending a gaze-value indication, wherein the gaze-value indication indicates the individual gaze value for the first user-account.

Abstract: Methods and systems for monitoring performance in a distributed storage system described. One example method includes identifying requests sent by clients to the distributed storage system, each request including request parameter values for request parameters; generating probe requests based on the identified requests, the probe requests including probe request parameter values for probe request parameter values, representing a statistical sample of the request parameters included in the identified requests; sending the generated probe requests to the distributed storage system over a network, wherein the distributed storage system is configured to perform preparations for servicing each probe request in response to receiving the probe request; receiving responses to the probe requests from the distributed storage system, and outputting at least one performance metric value measuring a current performance state of the distributed storage system based on the received responses.

Abstract: Provided is an in-ear device and associated computational support system that leverages machine learning to interpret sensor data descriptive of one or more in-ear phenomena during subvocalization by the user. An electronic device can receive sensor data generated by at least one sensor at least partially positioned within an ear of a user, wherein the sensor data was generated by the at least one sensor concurrently with the user subvocalizing a subvocalized utterance. The electronic device can then process the sensor data with a machine-learned subvocalization interpretation model to generate an interpretation of the subvocalized utterance as an output of the machine-learned subvocalization interpretation model.

Abstract: A method includes receiving an input text sequence to be synthesized into speech in a first language and obtaining a speaker embedding, the speaker embedding specifying specific voice characteristics of a target speaker for synthesizing the input text sequence into speech that clones a voice of the target speaker. The target speaker includes a native speaker of a second language different than the first language. The method also includes generating, using a text-to-speech (TTS) model, an output audio feature representation of the input text by processing the input text sequence and the speaker embedding. The output audio feature representation includes the voice characteristics of the target speaker specified by the speaker embedding.

Abstract: A method includes receiving a training example that includes audio data representing a spoken utterance and a ground truth transcription. For each word in the spoken utterance, the method also includes inserting a placeholder symbol before the respective word identifying a respective ground truth alignment for a beginning and an end of the respective word, determining a beginning word piece and an ending word piece, and generating a first constrained alignment for the beginning word piece and a second constrained alignment for the ending word piece. The first constrained alignment is aligned with the ground truth alignment for the beginning of the respective word and the second constrained alignment is aligned with the ground truth alignment for the ending of the respective word. The method also includes constraining an attention head of a second pass decoder by applying the first and second constrained alignments.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for transcoding and providing resources. A query is received. A resource is generated, and the resource, when processed by a client device, causes the client device to perform various operations. The operations include displaying a search result having an associated landing page and detecting user interaction with the search result. The operations include, in response to detecting user interaction with the search result, displaying a first portion of the landing page that is embedded within the resource and requesting, over a network, a second portion of the landing page. The operations include receiving the second portion of the landing page, and loading the second portion of the landing page. The generated resource is provided in response to the query.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining summary content for resources in a domain. In one aspect, a method includes accessing a first resource belonging to a particular domain, selecting an anchor in the first resource linking to a second resource belonging to the particular domain, identifying particular text content in the first resource that is subordinate to the anchor that the second resource includes the particular text content that is subordinate to the anchor, based on determining that the second resource includes the particular text content that is subordinate to the anchor, generating a domain template for the particular domain, the domain template specifying a location of the particular text content in the second resource, and determining, for each respective resource belonging to the particular domain having a structure matching the domain template, respective text content for the respective resource.

Abstract: A method for aligning a translation of original caption data with an audio portion of a video is provided. The method includes identifying, by a processing device, original caption data for a video that includes a plurality of caption character strings. The processing device identifies speech recognition data that includes a plurality of generated character strings and associated timing information for each generated character string. The processing device maps the plurality of caption character strings to the plurality of generated character strings using assigned values indicative of semantic similarities between character strings. The processing device assigns timing information to the individual caption character strings based on timing information of mapped individual generated character strings. The processing device aligns a translation of the original caption data with the audio portion of the video using assigned timing information of the individual caption character strings.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for detecting intermediate reinforcement learning goals. One of the methods includes obtaining a plurality of demonstration sequences, each of the demonstration sequences being a sequence of images of an environment while a respective instance of a reinforcement learning task is being performed; for each demonstration sequence, processing each image in the demonstration sequence through an image processing neural network to determine feature values for a respective set of features for the image; determining, from the demonstration sequences, a partitioning of the reinforcement learning task into a plurality of subtasks, wherein each image in each demonstration sequence is assigned to a respective subtask of the plurality of subtasks; and determining, from the feature values for the images in the demonstration sequences, a respective set of discriminative features for each of the plurality of subtasks.

Abstract: Cloud migration may be performed by identifying applications that are currently operating in the enterprise and performing certain determinations as to whether those applications are proper candidates for the migration to the cloud. One example method of operation may provide identifying at least one application operating on an enterprise network, retrieving current usage data of the at least one application, comparing the current usage data of the at least one application to a threshold amount of usage data to determine whether the application has exceeded the threshold amount of usage data. Next, the creation of an instance process may be performed on an entity operating outside the enterprise network and the application may be operated via the instance process and otherwise terminated in the enterprise network to alleviate resources.

Abstract: A method for automatic speech recognition using joint acoustic echo cancellation, speech enhancement, and voice separation includes receiving, at a contextual frontend processing model, input speech features corresponding to a target utterance. The method also includes receiving, at the contextual frontend processing model, at least one of a reference audio signal, a contextual noise signal including noise prior to the target utterance, or a speaker embedding including voice characteristics of a target speaker that spoke the target utterance. The method further includes processing, using the contextual frontend processing model, the input speech features and the at least one of the reference audio signal, the contextual noise signal, or the speaker embedding vector to generate enhanced speech features.

Abstract: Techniques and apparatuses are described for user equipment-coordination set, UECS, hybrid automatic repeat request, HARQ, that establish a HARQ timeline that is specific to the capabilities of a respective UECS. Compared to a single user equipment, UE, the HARQ timeline for a UECS depends on a number of factors, such as the joint processing capability in the UECS, latency of communication over a local wireless network between the UEs in the UECS, or the like. Based on its capabilities, the UECS can request uplink and/or downlink processing delay times or a UECS-specific HARQ timeline from a base station. The base station grants the uplink and/or downlink processing delay times or the UECS-specific HARQ timeline to the UECS in a layer-1, layer-2, or a layer-3 control message. The use of a UECS-specific HARQ timeline increases the reliability of HARQ signaling for uplink and downlink communication between the UECS and a base station.

Abstract: A method for principal component analysis includes receiving a principal component analysis (PCA) request from a user requesting data processing hardware to perform PCA on a dataset, the dataset including a plurality of input features. The method further includes training a PCA model on the plurality of input features of the dataset. The method includes determining, using the trained PCA model, one or more principal components of the dataset. The method also includes generating, based on the plurality of input features and the one or more principal components, one or more embedded features of the dataset. The method includes returning the one or more embedded features to the user.

Abstract: Techniques and apparatuses are described for joint-processing of random access channel communications to improve the reliability and/or the geographic range of a random access procedure for a user equipment. Joint-processing, including joint-transmission and/or joint-reception, by a user equipment-coordination set on behalf of a single user equipment, or by an Active Coordination Set of base stations with the single user equipment, can improve the link budget of Random Access Channel communications and facilitate network access for a user equipment at a greater distance from a base station or in the face of challenging channel conditions.

Abstract: Implementations described herein relate to methods, systems, and computer-readable media to automatically answer a call. In some implementations, a method includes receiving a call from a caller device at a client device. The method further includes determining, based on an identifier associated with the call, whether the call matches auto answer criteria, and yin response to determining that the call matches the auto answer criteria, answering the call without user input and without alerting a user of the client device. The method further includes generating a call embedding for the call based on received audio of the call, comparing the call embedding with spam embeddings to determine whether the call is a spam call, and in response to determining that the call is a spam call, terminating the call.

Abstract: Techniques and apparatuses are described for secondary cell-user equipment (SC-UE) handovers. In aspects, a base station provides, as a primary cell, primary cell services to a first group of user equipments (UEs) in a first base station-user equipment dual connectivity (BUDC) group. In implementations, the base station determines to perform an SC-UE handover that disconnects a first UE from a first secondary cell-user equipment (SC-UE) that provides a first secondary cell to the first BUDC group and connects the first UE to a second SC-UE that provides a second secondary cell to a second BUDC group. The base station directs the first SC-UE to release the first UE from the first BUDC group, and communicates, as the primary cell, control-plane information or user-plane data with one or more UEs in the first group of UEs that remain in the first BUDC group.

Abstract: A method for scalable event driven auto-diagnosis systems includes obtaining a data packet configured for transmission across a network from a source address to a destination address. The method includes obtaining a list of changes to the network. The method also includes analyzing, based on a network model, the data packet using a plurality of analyzers. The method includes correlating the list of changes to the network and the analysis of the data packet. The method further includes determining, based on the correlation between the list of changes to the network and the analysis of the data packet, a configuration status of the network. The method also includes reporting the configuration status to a user.

Abstract: A method includes receiving a voice request to generate synthesized voiceover speech for a target advertisement having one or more advertising campaign attributes. The method also includes generating, based on the one or more advertising campaign attributes, a voiceover script that includes a sequence of text for the synthesized voiceover speech. The method also includes generating, using a text-to-speech (TTS) system, the synthesized voiceover speech. The TTS system is configured to receive, as input, the sequence of text for the voiceover script and generate, as output, the synthesized voiceover speech. Here, the synthesized voiceover speech has speech characteristics specified by a target TTS vertical. The method also includes overlaying the synthesized voiceover speech on the target advertisement.

Abstract: A method for dynamically changing a graphical user interface element occurs in response to detecting that a temporal user interface element displayed on a user interface of user device. The method includes receiving, at the user device, a contextual signal characterizing a state of a user. The method further includes determining, by the user device, that the contextual signal characterizing the state of the user is indicative of the user intending to interact with a temporal user interface element. The methods also include, in response to determining that the contextual signal characterizing the state of the user is indicative of the user intending to interact with a temporal user interface element, modifying a respective state of the temporal user interface element displayed on the user interface of the user device.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for rollback resistant security are disclosed. In one aspect, a method, during a boot process of a computing device, includes the actions of obtaining a secret key derived from device-specific information for the computing device. The method further includes verifying that a signature for a software module is valid. The method further includes obtaining information indicating a current version of the software module. The method further includes using the secret key to generate a first encryption key corresponding to the current version of the software module and a second encryption key corresponding to a prior version of the software module. The method further includes preventing future access to the secret key until the computing device is rebooted. The method further includes providing the software module access to the first encryption key and the second encryption key.