Abstract: An aspect of the disclosed technology comprises a test-time adaptation (“TTA”) technique for visual document understanding (“VDU”) tasks that uses self-supervised learning on different modalities (e.g., text and layout) by applying masked visual language modeling (“MVLM”) along with pseudo-labeling. In accordance with an aspect of the disclosed technology, the TTA technique enables a document model to adapt to domain or distribution shifts that are detected.

Abstract: Techniques and devices for communicating Internet Protocol version 6 (IPv6) traffic using a lightweight Internet Protocol (LwIP) network stack and a tunnel router, by an electronic device in a first network segment of a fabric network are described. The tunnel router receives an Internet Protocol version (IPv4) address for an access router in a second network segment of the fabric network and establishes a secure IPv6-over-IPv4 network tunnel with the access router in the second network segment. The tunnel router advertises an IPv6 network route for the fabric network and, using the advertised IPv6 network route, routes IPv6 messages from the electronic device to one or more devices in the second network segment using the secure IPv6-over-IPv4 network tunnel.

Abstract: A method for verifying structured data includes receiving structured data, deconstructing the structured data into corresponding elements and obtaining standard structured data having corresponding standard elements. The method also includes comparing the elements of the structured data with the standard elements of the standard structured data to identify any element differences. For each element difference, the method includes comparing the element difference against a registry of element comparisons, determining whether the element difference is expected or unexpected based on a heuristic or at least one rule, and when the element difference is unexpected, generating a signal indicating the presence of an unexpected element in the structured data.

Abstract: In one general aspect, a method can include displaying, on a display device included in a computing device, content in an application executing on the computing device. The method can further include displaying, in a user interface on the display device, at least one identifier, receiving a selection of the at least one identifier, and initiating casting in response to receiving the selection of the at least one identifier.

Abstract: A method includes receiving image data for a plurality of objects of interest of a user, and receiving a query from the user referring to the plurality of objects and requesting a digital assistant to a discern color matching insight associated with the plurality of objects referred to by the query. The method includes processing the query and the image data to: identify, for each particular object of the plurality of objects referred to by the query, one or more corresponding colors of the particular object; and determine, based on the corresponding colors, the color matching insight associated with the plurality objects for the digital assistant to discern. The method includes generating, for output from a user device associated with the user, content indicating the discerned color matching insight associated with the plurality of objects.

Abstract: This document describes methods and systems of on-device real-time translation for media content on a mobile electronic device. The translation is managed and executed by an operating system of the electronic device rather than within a particular application executing on the electronic device. The operating system can translate media content, including visual content displayed on a display device of the electronic device or audio content output by the electronic device. Because the translation is at the OS level, the translation can be implemented, automatically or based on a user input, across a variety of (including all) applications and a variety of content on the electronic device to provide a consistent translation experience, which is provided via a system UI overlay that displays translated text as captions to video content or as a replacement to on-screen text.

Abstract: The present disclosure provides systems and methods for activating a general purpose input/out (“GPIO”) pin. The system may include a control device, agency device, and a controlled device. The agency device may be part of the controlled device or it may be a separate device. The agency device may receive a communication signal transmitted from the control device. The agency device may count the number of pulses during a certain period of time to determine a GPIO pin to trigger on the controlled device.

Abstract: Techniques and apparatuses are described that implement a metal structure for steering and broadening millimeter-wave (mmWave) antenna coverage. A user device includes at least one mmWave antenna that produces a near-field radiation region and a far-field radiation pattern. Disposed within the near-field radiation region is the metal structure comprising one or more metal pieces. The metal pieces are able to reflect energy associated with the far-field radiation pattern or produce another far-field radiation pattern based on currents induced in the metal pieces by the near-field radiation region in such a way that the far-field radiation pattern from the mmWave antenna is positively affected (e.g., steered and/or broadened). In this way, the far-field radiation pattern can be effectively steered and/or broadened with a simple cost and space-effective design.

Abstract: This document describes systems and techniques for simulating the touch of a human finger in manipulating an interface device, such as a touchscreen included in a mobile phone or other computing device. The systems and techniques include an artificial finger configured to be received and manipulated by a robotic actuator to simulate surface engagement, mechanical force, and electrical conductivity of a human finger engaging the touchscreen at varied touch angles and/or mechanical pressures. The systems and techniques thereby provide for rigorous and repeatable testing of an electrical and mechanical response of the touchscreen to simulated user inputs without involving a human test operator.

Abstract: A method for training a speech recognition model with a minimum word error rate loss function includes receiving a training example comprising a proper noun and generating a plurality of hypotheses corresponding to the training example. Each hypothesis of the plurality of hypotheses represents the proper noun and includes a corresponding probability that indicates a likelihood that the hypothesis represents the proper noun. The method also includes determining that the corresponding probability associated with one of the plurality of hypotheses satisfies a penalty criteria. The penalty criteria indicating that the corresponding probability satisfies a probability threshold, and the associated hypothesis incorrectly represents the proper noun. The method also includes applying a penalty to the minimum word error rate loss function.

Abstract: Techniques and apparatuses are described for enhanced uplink spectrum sharing. A base station communicates, to a user equipment (UE) an air interface resource configuration for a second air interface resource allocated to a second cell that uses a second radio access technology (RAT) and implemented by the base station. The base station receives a first air interface resource configuration for a first air interface resource allocated to a first cell that uses a first RAT, where the first air interface resource configuration differs from the second air interface resource configuration. In aspects, the base station communicates the first air interface resource configuration to the UE. Based on receiving a low-utilization indication for the first air interface resource, the base station directs the UE to utilize the first air interface resource for transmitting uplink communications using the second RAT.

Abstract: Techniques and apparatuses are described that implement collocated mm Wave and sub-6 GHz antennas. An apparatus includes at least one mmWave antenna that produces a near-field radiation region and a far-field radiation pattern in a mmWave frequency band. Disposed within the near-field radiation region is a sub-6 GHz antenna that produces a radiation pattern in a sub-6 GHz frequency band. The sub 6 GHz antenna is able to positively affect the far-field radiation pattern from the mm Wave antenna (e.g., via steering and/or broadening). In this way, the mmWave antenna and the sub-6 GHz antenna can be collocated to conserve space while also steering and/or broadening the far-field radiation pattern of the mm Wave antenna.

Abstract: This document discloses systems and methods for implementing automatic test parameter tuning in constrained random verification. In aspects, a method receives a first set of parameters for testing a design under test, performs a first regression (e.g., an overnight regression test) on a design under test using the first set of parameters, and analyzes the results of the first regression including determining a coverage percentage. The method then generates an optimized set of parameters based on the analysis of the results of the first regression and performs an additional regression on the design under test using the optimized set of parameters. In aspects, the method is repeated using the optimized set of parameters until a coverage percentage is reached, or in some implementations, full coverage may be reached. Some implementations of the method utilize black-box optimization through use of a Bayesian optimization algorithm.

Abstract: A computer-implemented method includes receiving, from a user account, an operation request to perform an operation on data stored at a distributed computing environment. The operation request includes an on-premises token associated with the user account. The method also includes extracting, from the on-premises token, an access scope associated with permissions of the user account for accessing the data stored at the distributed computing environment. The method also includes translating the extracted access scope into query parameters compatible with the data stored at the distributed computing environment. The method also includes determining, using the query parameters, whether the operation request can access the data stored at the distributed computing environment. When the operation request can access the data stored at the distributed computing environment, the method includes executing the operation request.

Abstract: A computer-implemented method for generating a proposed bid includes receiving an input generated by a first content provider as part of a bidding process where content providers bid for opportunities to provide content for publication. The input indicates (i) a resource for publishing first content from the first content provider and (ii) a campaign characteristic associated with the first content. The method includes determining a proposed bid for the first content provider based on the input and on historical data from the bidding process associated with the resource. The method includes presenting the proposed bid to the first content provider.

Abstract: Physical gestures may be performed by a user to interact with a head-mounted device. These physical gestures can be detected by other devices worn by the user. For example, a fitness tracker worn on a wrist of a user may be used to detect the movement of the wrist as a gesture and then communicate this gesture to the head-mounted device. This accessory-mediated gesture interaction may be interrupted when the accessory device loses power or otherwise stops tracking. In this condition, the disclosed head-mounted device may take-over gesture detection by monitoring distortions to a magnetic field caused by movement of the device. This form of gesture detection can be used when an active device has exhausted its battery. Further, this form of gesture detection can be used for electrically inactive objects that can cause detectable changes to the magnetic field when moved in the magnetic field.

Abstract: A method for training a non-autoregressive TTS model includes obtaining a sequence representation of an encoded text sequence concatenated with a variational embedding. The method also includes using a duration model network to predict a phoneme duration for each phoneme represented by the encoded text sequence. Based on the predicted phoneme durations, the method also includes learning an interval representation and an auxiliary attention context representation. The method also includes upsampling, using the interval representation and the auxiliary attention context representation, the sequence representation into an upsampled output specifying a number of frames. The method also includes generating, based on the upsampled output, one or more predicted mel-frequency spectrogram sequences for the encoded text sequence.

Abstract: The technology provides for a system for determining a gesture provided by a user. In this regard, one or more processors of the system may receive image data from one or more visual sensors of the system capturing a motion of the user, and may receive motion data from one or more wearable computing devices worn by the user. The one or more processors may recognize, based on the image data, a portion of the user's body that corresponds to a gesture to perform a command. The one or more processors may also determine one or more correlations between the image data and the received motion data. Based on the recognized portion of the user's body and the one or more correlations between the image data and the received motion data, the one or more processors may detect the gesture.

Abstract: A method includes obtaining a set of training samples. During each of a plurality of training iterations, the method also includes sampling a batch of training samples from the set of training samples. The method includes, for each training sample in the batch of training samples, determining, using a data value estimator, a selection probability. The selection probability for the training sample is based on estimator parameter values of the data value estimator. The method also includes selecting, based on the selection probabilities of each training sample, a subset of training samples from the batch of training samples, and determining, using a predictor model with the subset of training samples, performance measurements. The method also includes adjusting model parameter values of the predictor model based on the performance measurements, and updating the estimator parameter values of the data value estimator based on the performance measurements.

Abstract: A method of accessing data includes storing a table that includes a plurality of tablets corresponding to distinct non-overlapping table portions. Respective pluralities of tablet access objects and application objects are stored in a plurality of servers. A distinct application object and distinct tablet are associated with each tablet access object. Each application object corresponds to a distinct instantiation of an application associated with the table. The tablet access objects and associated application objects are redistributed among the servers in accordance with a first load-balancing criterion. A first request directed to a respective tablet is received from a client. In response, the tablet access object associated with the respective tablet is used to perform a data access operation on the respective tablet, and the application object associated with the respective tablet is used to perform an additional computational operation to produce a result to be returned to the client.