Abstract: Techniques are disclosed herein for focused training of language models and end-to-end hypertuning of the framework. In one aspect, a method is provided that includes obtaining a machine learning model pre-trained for language modeling, and post-training the machine learning model for various tasks to generate a focused machine learning model. The post-training includes: (i) training the machine learning model on an unlabeled set of training data pertaining to a task that the machine learning model was pre-trained for as part of the language modeling, and the unlabeled set of training data is obtained with respect to a target domain, a target task, or a target language, and (ii) training the machine learning model on a labeled set of training data that pertains to another task that is an auxiliary task related to a downstream task to be performed using the machine learning model or output from the machine learning model.

Abstract: Disclosed herein are embodiments of systems and methods for stable and elevated idle-mode temperature for assembled semiconductor devices. In an embodiment, a processor includes a communication interface configured to receive, from a first hardware component, instructions assigned to the processor for execution. The processor also includes temperature-measurement circuitry configured to monitor an on-chip temperature of the processor. The processor also includes control logic configured to: determine whether the processor is active or idle; determine whether the on-chip temperature of the processor exceeds a first threshold; based on determining that the processor is idle and that the on-chip temperature of the processor exceeds the first threshold, disable one or more idle-mode power-saving features of the processor; and selectively adjust one or more operating parameters of the processor to keep the on-chip temperature of the processor between the first threshold and a second (higher) threshold.

Abstract: A docking station for use with a solar tracking system includes a frame configured to selectively support solar cleaning equipment thereon and a mounting bracket operably coupled to a portion of the frame. The mounting bracket maintains a gap between adjacent edges of the frame and an adjacent solar module, and the frame includes a width that approximates a width of the solar module.

Abstract: Exemplary systems and methods are disclosed for allocating resources in a breeding pipeline to multiple origins. One exemplary method includes accessing a data structure including data representative of multiple origins, in which the data includes, for each of the multiple origins, a trait performance expression or genotypic component information. The exemplary method further includes determining a resource allocation, which allocates n resources among the multiple origins based on a probability associated with the trait performance expressions and/or the genotypic components for the origins, and then allocating the n resources in the breeding pipeline for the multiple origins, based on the determined resource allocation.

Abstract: An Automated Parking System (“APS”) for a motor vehicle includes a plurality of cameras for generating a vision signal, in response to the cameras capturing an image of a region surrounding the motor vehicle. The APS includes a processor for receiving the vision signal from the cameras. The APS further includes a non-transitory computer readable storage medium for storing instructions such that the processor is programmed to execute a plurality of routines. The routines include a detection module for detecting one or more landmark points in the image. The routines further include an estimate slot module for determining one or more corners of a parking slot based on the landmark points. The routines further include a maneuver module for generating an action signal, with a power steering system maneuvering the motor vehicle into the parking slot based on the corners.

Abstract: Various methods are provided for facilitating heterogeneous database migration. One example method may comprise accessing a migration configuration file, the migration configuration file comprising information indicative of at least a source database type of a source database, connection information of accessing the source database, a destination database type of a destination database, and connection information for accessing the destination database, generating configuration data based on the source database type and the destination database type to generate a mapping of a source database table stored within the source database to a destination database table stored within the destination database, and generating a query for execution on the source database enabling data migration from the source database to the destination database.

Abstract: In an approach, a processor receives a request to access an electronic resource from a device. A processor causes the device to generate a Completely Automated Public Turing test (CAPTCHA), where the CAPTCHA comprises: a virtual keyboard; an ordered string of characters required to be input; and presentation of a highlighted key of the virtual keyboard on the device, wherein (i) the highlighted key is a first key visually distinct from other keys of the virtual keyboard and (ii) the first key corresponds to a character of the ordered string of characters. A processor receives a result of the CAPTCHA from the device. A processor performs an action based on the result.

Abstract: A clamp assembly for use with a solar module includes a first clamping arm operably coupled to a first portion of a solar module, a second clamping arm operably coupled to a second portion of the solar module and operably coupled to the first clamping arm and disposed in juxtaposed relation thereto, and a fastener operably coupled to a respective portion of each of the first and second clamping arms. The fastener is configured to draw each of the first and second clamping arms towards one another from a first, open position, where the solar module is free to move relative to first and second clamping arms to a second, closed position, where the solar module is inhibited from moving relative to the first and second clamping arms.

Abstract: In an embodiment, a processor includes at least one core and power management logic. The power management logic is to receive temperature data from a plurality of dies within a package that includes the processor, and determine a smallest temperature control margin of a plurality of temperature control margins. Each temperature control margin is to be determined based on a respective thermal control temperature associated with the die and also based on respective temperature data associated with the die. The power management logic is also to generate a thermal report that is to include the smallest temperature control margin, and to store the thermal report. Other embodiments are described and claimed.

Abstract: The present disclosure relates to techniques for using variant inconsistency attack (VIA) as a simple and effective adversarial attack method to create useful adversarial examples for adversarial training of machine-learning models. In one particular aspect, a method is provided that includes obtaining a set of input examples for attacking a machine-learning model (the set of examples do not have corresponding labels), modifying an example from the set of examples in a utility preserving manner to generate a pair of modified examples, attacking the machine-learning model with the pair of modified examples in order generate a pair of predictions for the pair of modified examples, comparing the pair of predictions to determine whether the pair of predictions are the same or different, and in response to the pair of predictions being different, adding the pair of modified examples to a set of adversarial examples.

Abstract: A profile associated with a user that has unintentionally disconnected from an electronic conference is determined. Content missed by the user is determined. A determination is made whether the content missed by the user is relevant to the user based on the determined profile associated with the user and an indicated preference of the user. In response to determining that the content missed by the user is relevant to the user, the content missed by the user is transmitted to a computing device of the user.

Abstract: Aspects of the invention perform an operation comprising receiving a plurality of rules associated with a first seat in a mass-transit vehicle, determining a plurality of physical attributes for a first passenger, of a plurality of passengers, based on data describing the first passenger received from a plurality of data sources, determining a plurality of non-physical attributes for the first passenger based on the received data describing the first passenger, determining that at least one of the plurality of physical attributes and the plurality of non-physical attributes of the first passenger satisfy each rule in the set of rules associated with the first seat, computing, based on a machine learning (ML) model a score for the first passenger, determining that the score exceeds a threshold score associated with the first seat, and allocating the first passenger to the first seat in the mass-transit vehicle.

Abstract: An embodiment includes activating, responsive to receiving an update notification, an update mode of a mobile device, wherein the activating of the update mode includes disabling a primary communication interface and enabling a secondary communication interface, and wherein the update notification includes notification of a software update available for the mobile device. The embodiment also includes initiating execution of the software update on the mobile device while the mobile device remains in the update mode. The embodiment also includes deactivating, responsive to completing the software update, the update mode of the mobile device, wherein the deactivating of the update mode includes enabling the primary communication interface and disabling the secondary communication interface.

Abstract: A profile associated with a user that has unintentionally disconnected from an electronic conference is determined. Content missed by the user is determined. A determination is made whether the content missed by the user is relevant to the user based on the determined profile associated with the user and an indicated preference of the user. In response to determining that the content missed by the user is relevant to the user, the content missed by the user is transmitted to a computing device of the user.

Abstract: Aspects of the present invention disclose a method for performing playback functions of a multimedia feed based on a task of a request and an activity of a user. The method includes one or more processors identifying a task of multimedia of a video sharing application of a computing device. The method further includes determining an activity of a user utilizing a live video feed of a camera of the computing device. The method further includes determining whether the activity of the user correlates to the task of the multimedia. The method further includes in response to determining that the activity of the user correlates to the task of the multimedia, initiating a playback function of the video sharing application of the computing device based at least in part on the activity of the user and the task of the multimedia.

Abstract: Various methods are provided for facilitating heterogeneous database migration. One example method may comprise accessing a migration configuration file, the migration configuration file comprising information indicative of at least a source database type of a source database, connection information of accessing the source database, a destination database type of a destination database, and connection information for accessing the destination database, generating configuration data based on the source database type and the destination database type to generate a mapping of a source database table stored within the source database to a destination database table stored within the destination database, and generating a query for execution on the source database enabling data migration from the source database to the destination database.

Abstract: A system includes multiple processors and a power controller. Each processor includes a throttling engine. The power controller is to, in response to a determination that a first power consumption level exceeds a first threshold, assert a critical signal to each throttling engine of the plurality of processors. Further, for each processor, the throttling engine of the processor is to perform a sequence of multiple throttling states while the critical signal is asserted by the power controller, where the sequence of multiple throttling states is performed according to a state machine of the throttling engine. Other embodiments are described and claimed.

Abstract: Aspects of the invention perform an operation comprising receiving a plurality of rules associated with a first seat in a mass-transit vehicle, determining a plurality of physical attributes for a first passenger, of a plurality of passengers, based on data describing the first passenger received from a plurality of data sources, determining a plurality of non-physical attributes for the first passenger based on the received data describing the first passenger, determining that at least one of the plurality of physical attributes and the plurality of non-physical attributes of the first passenger satisfy each rule in the set of rules associated with the first seat, computing, based on a machine learning (ML) model a score for the first passenger, determining that the score exceeds a threshold score associated with the first seat, and allocating the first passenger to the first seat in the mass-transit vehicle.

Abstract: Techniques are provided for using context tags in named-entity recognition (NER) models. In one particular aspect, a method is provided that includes receiving an utterance, generating embeddings for words of the utterance, generating a regular expression and gazetteer feature vector for the utterance, generating a context tag distribution feature vector for the utterance, concatenating or interpolating the embeddings with the regular expression and gazetteer feature vector and the context tag distribution feature vector to generate a set of feature vectors, generating an encoded form of the utterance based on the set of feature vectors, generating log-probabilities based on the encoded form of the utterance, and identifying one or more constraints for the utterance.

Abstract: Techniques are disclosed for systems including techniques for multi-feature balancing for natural langue processors. In an embodiment, a method includes receiving a natural language query to be processed by a machine learning model, the machine learning model utilizing a dataset of natural language phrases for processing natural language queries, determining, based on the machine learning model and the natural language query, a feature dropout value, generating, and based on the natural language query, one or more contextual features and one or more expressional features that may be input to the machine learning model, modifying at least one or the one or more contextual features and the one or more expressional features based on the feature dropout value to generate a set of input features for the machine learning model, and processing the set of input features to cause generating an output dataset for corresponding to the natural language query.