Abstract: A method (300) and apparatus (110, 150) collect and upload implicit analytic data. The method can include storing (320) dependency rules corresponding explicit events to implicit events. The method can include collecting (330) and storing (340) implicit event data corresponding to implicit events. The method can include receiving (350) an explicit event at the device. The method can include evaluating (360) dependency rules corresponding to the explicit event. The method can include identifying (370) a relevant subset of implicit event data corresponding to the explicit event based on evaluating the dependency rules. The method can include uploading (390) the relevant subset of the implicit event data and explicit event data corresponding to the explicit event.

Abstract: A method for optimizing speech recognition includes receiving a first acoustic segment characterizing a hotword detected by a hotword detector in streaming audio captured by a user device, extracting one or more hotword attributes from the first acoustic segment, and adjusting, based on the one or more hotword attributes extracted from the first acoustic segment, one or more speech recognition parameters of an automated speech recognition (ASR) model. After adjusting the speech recognition parameters of the ASR model, the method also includes processing, using the ASR model, a second acoustic segment to generate a speech recognition result. The second acoustic segment characterizes a spoken query/command that follows the first acoustic segment in the streaming audio captured by the user device.

Abstract: This document describes a speaker-integration system and associated devices and systems. The speaker-integration system includes a speaker that is re-workably mounted, via an elastomeric gasket, to an intermediate structure that is mountable to an outer enclosure of an electronic device. The intermediate structure forms a cavity in which the speaker is sealed, effective to use the cavity as the speaker's back volume to contain acoustic waves without impacting other structures in the electronic device. The front of the speaker is sealed against the outer enclosure by a gasket that controls, based on its placement and geometry, axial and radial directions of the speaker relative to the intermediate structure to prevent the speaker from buzzing against surrounding rigid parts. The speaker has wires that exit the back volume via a detachable grommet, which controls positioning of the wires to prevent rub and buzz against surrounding parts.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for computing a layer output for a convolutional neural network layer, the method comprising: receiving the layer input, the layer input comprising a plurality of activation inputs, the plurality of activation inputs represented as a multi-dimensional matrix comprising a plurality of depth levels, each depth level being a respective matrix of distinct activation inputs from the plurality of activation inputs; sending each respective kernel matrix structure to a distinct cell along a first dimension of the systolic array; for each depth level, sending the respective matrix of distinct activation inputs to a distinct cell along a second dimension of the systolic array; causing the systolic array to generate an accumulated output from the respective matrices sent to the cells; and generating the layer output from the accumulated output.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, that creates a machine learning model with k-means clustering. In some implementations, an instruction to create a model is obtained. A data set including geographic data and non-geographic data is received. The data set includes multiple data entries. Geographic centroids are determined from the geographic data. The data set is analyzed to obtain statistics of the data set. Transformed data is generated from the data set, the statistics, and the geographic centroids. A model is generated with the transformed data, the model indicating multiple data groupings.

Abstract: A method for pre-utilizing tables before refresh includes selecting a query for execution in a data analytics application and identifying a table specified by the query, for instance a materialized table. Another operation includes computing an encoding, such as a hash value, for at least one portion of the identified table. Another operation includes comparing the computed encoding to a pre-stored encoding for the at least one portion of the identified table. On the condition that the computed encoding is identical to the pre-stored encoding, the operations include utilizing the identified table in executing the query in lieu of re-materializing the identified table. Otherwise, on the condition that the computed encoding differs from the pre-stored encoding, the operations include re-materializing the identified table before returning the results of the query.

Abstract: This document includes techniques, apparatuses, and systems related to an interface for revision-limited memory, which can improve various computing aspects and performance. In aspects, confidentiality, integrity, and availability may be ensured while increasing the performance of revision-limited memory. In this example, the techniques also enable the digital computing device to interact with information related to the revision-limited memory.

Abstract: A method includes receiving a model analysis request from a user. The model analysis requests requesting the data processing hardware to provide one or more statistics of a model trained on a dataset. The method also includes obtaining the trained model. The trained model includes a plurality of weights. Each weight is assigned to a feature of the trained model. The model also includes determining, using the dataset and the plurality of weights, the one or more statistics of the trained model based on a linear regression of the trained model. The method includes reporting the one or more statistics of the trained model to the user.

Abstract: A method includes receiving a plurality of unlabeled audio samples corresponding to spoken utterances not paired with corresponding transcriptions. At a target branch of a contrastive Siamese network, the method also includes generating a sequence of encoder outputs for the plurality of unlabeled audio samples and modifying time characteristics of the encoder outputs to generate a sequence of target branch outputs. At an augmentation branch of a contrastive Siamese network, the method also includes performing augmentation on the unlabeled audio samples, generating a sequence of augmented encoder outputs for the augmented unlabeled audio samples, and generating predictions of the sequence of target branch outputs generated at the target branch. The method also includes determining an unsupervised loss term based on target branch outputs and predictions of the sequence of target branch outputs. The method also includes updating parameters of the audio encoder based on the unsupervised loss term.

Abstract: Techniques and apparatuses are described for reducing a flicker effect of multiple light sources in an image captured with an imaging device. A lighting frequency associated with each of the multiple light sources is detected and prioritized relative to a flicker effect upon the image to identify at least a first-prioritized lighting frequency and a second-prioritized lighting frequency. A first exposure-time factorization set is determined for the first-prioritized lighting frequency, and a second exposure-time factorization set is determined for the second-prioritized lighting frequency. An exposure time of the imaging device is adjusted to an exposure time identified in the first exposure-time factorization set that matches, or aligns near-to-matching, an exposure time identified in the second exposure-time factorization set.

Abstract: A method (400) includes receiving, at a first processor (110) of a user device (102), streaming multi-channel audio (118) captured by an array of microphones (107), each channel (119) including respective audio features. For each channel, the method also includes processing, by the first processor, using a first stage hotword detector (210), the respective audio features to determine whether a hotword is detected. When the first stage hotword detector detects the hotword, the method also includes the first processor providing chomped raw audio data (212) to a second processor that processes, using a first noise cleaning algorithm (250), the chomped raw audio data to generate a clean monophonic audio chomp (260). The method also includes processing, by the second processor using a second stage hotword detector (220), the clean monophonic audio chomp to detect the hotword.

Abstract: Described herein is a method for managing media item playback. A method includes presenting a media player in a first portion of a user interface (UI) provided by a first application on a screen of a user device, the media player to provide playback of a first media item from a first post of a plurality of posts of the first application. The method also includes relocating the media player to a second portion of the UI while continuing to provide playback of the first media item by the relocated media player, and receiving a user gesture to scroll through the plurality of posts of the first application on the screen of the user device, the plurality of posts comprising a second post having a second media item.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating subqueries from a query. In one aspect, a method includes obtaining a query, generating a set of two subqueries from the query, where the set includes a first subquery and a second subquery, determining a quality score for the set of two subqueries, determining whether the quality score for the set of two subqueries satisfies a quality threshold, and in response to determining that the quality score for the set of two subqueries satisfies the quality threshold, providing a first response to the first subquery that is responsive to a first operation that receives the first subquery as input and providing a second response to the second subquery that is responsive to a second operation that receives the second subquery as input.

Abstract: An access system for a wireless local area network is provided. The access system initiates operation in a first frequency band of a plurality of frequency bands of the wireless local area network to provide one or more wireless client devices of the wireless local area network with access to a wireless wide area network. The access system also processes one or more association requests received in the first frequency band to identify one or more associated wireless client device. The access system also determines whether each of the one or more associated wireless client devices supports a second frequency band. The access system also sends a request to the one or more associated wireless client devices to transition from the first frequency band to the second frequency band based on determining that each of the one or more associated wireless client devices supports the second frequency band.

Abstract: A magnetic adapter for a cable connector includes a main body having an opening, a compression surface exposed within the opening and configured to compress a biasing retention clip of the cable connector, at least one locking surface configured to secure the biasing retention clip in a second non-locking position; and at least one magnet adjacent the opening. When the biasing retention clip is positioned within the opening, the compression surface causes the biasing retention clip to move from a first locking position where the retention clip is in a fully biased position to the second non-locking position where the retention clip is compressed.

Abstract: A method for database syncing is provided. The method includes receiving database change indicators indicating changes to a database. For each database change indicator, the method also includes determining whether the respective change to the database corresponds to a single change to a single version of the database, one of multiple changes to a common version of the database, or one of multiple changes to different versions of the database. The method further includes generating a change log compaction data structure comprising entries. Each entry includes an exact entry defining a corresponding single change to a corresponding single version of the database, a range entry defining corresponding multiple changes to a corresponding common version of the database, or an interval entry defining corresponding multiple changes to corresponding different versions of the database.

Abstract: Systems, articles, and methods integrate photopolymer film with eyeglass lenses. One or more hologram(s) may be recorded into/onto the photopolymer file to enable the lens to be used as a transparent holographic combiner in a wearable heads-up display employing an image source, such as a microdisplay or a scanning laser projector. The methods of integrating photopolymer film with eyeglass lenses include: positioning photopolymer film in a lens mold and casting the lends around the photopolymer film; sandwiching photopolymer film in between two portions of a lens applying photo polymer film to a concave surface of a lens and/or affixing a planar carrier (with photopolymer film thereon) to two points across a length of a concave surface of a lens.

Abstract: An apparatus includes a camera, a primary display panel including a pixel array and an aperture adjacent the pixel array, an auxiliary display, and an optical assembly including a reflecting optical element and an actuator coupled to the reflecting optical element. The actuator is configured to switch the reflecting optical element between a first arrangement and a second arrangement. The first arrangement defines an optical path from the aperture to the camera and the second arrangement defines an optical path from the aperture to the auxiliary display.

Abstract: A method includes obtaining a plurality of training data sets each associated with a respective native language and includes a plurality of respective training data samples. For each respective training data sample of each training data set in the respective native language, the method includes transliterating the corresponding transcription in the respective native script into corresponding transliterated text representing the respective native language of the corresponding audio in a target script and associating the corresponding transliterated text in the target script with the corresponding audio in the respective native language to generate a respective normalized training data sample.