Abstract: Questions play a central role in assessment of a candidate's expertise during an interview or examination. However, generating such questions from input text documents manually needs specialized expertise and experience. Further, techniques that are available for automated question generation require input sentence as well as an answer phrase in that sentence to generate question. This in-turn requires large training datasets consisting tuples of input sentence answer-phrase and the corresponding question. Additionally, training datasets are available are for general purpose text, but not for technical text. Present application provides systems and methods for generating technical questions from technical documents. The system extracts meta information and linguistic information of text data present in technical documents. The system then identifies relationships that exist in provided text data. The system further creates one or more graphs based on the identified relationships.

Abstract: The present disclosure is related in general to field of machine learning and image processing, that provides a method and system for detecting lane pattern. A lane classification system receives an input image comprising lane markings and co-ordinates of the lane markings, which is divided into two portions, a first image portion and a second image portion based on the lane markings and the co-ordinates of the lane markings. Further, the first image portion and the second image portion are resized into a predefined template size. The lane classification system detects a lane pattern in the two portions by analysing each of the one or more lane markings located in the resized first image portion and the second image portion, independently, based on historical lane data.

Abstract: A text recognition system causes a trained region encoder to determine a region of interest of an image file. The system modifies a first image associated with the first region of interest (e.g., parsed out from the first region) to generate a data augmentation entity that includes a modified image. Using a trained instance encoder, the system generates a first set of visual instances corresponding to the first region of interest image and a second set of visual instances corresponding to the data augmentation entity. The system generates the corresponding first and second sequences. By executing a self-supervised contrastive loss function on the first and second sequences, the system automatically updates a continual knowledge distillation model of the trained region encoder. The system provides the first sequence to an instance decoder to generate output text in response to the prompt.

Abstract: The present disclosure relates to a method and apparatus for generating training data of a deep learning model for lane classification. The method according to an embodiment of the present disclosure is performed by an electronic apparatus, and is a method for generating training data of a deep learning model for lane classification by generating a composite image of the other color lane using images of a white lane and the other color lane, and includes determining a ratio of other two channels based on one channel (reference color channel) for three color channels of red (R), green (G) and blue (B) of the other color lane in the image of the other color lane; and generating a composite image of the other color lane by scaling the image of the white lane by applying the determined ratio to the other two channels with respect to the reference color channel of the white lane.

Abstract: State of art techniques mostly rely of computationally intensive, time consuming Neural Networks. Embodiments provide a method and system for identification and classification of different grain and adulterant types for grain grading analysis. The method analyzes input image of grain sample of elements to determine morphological features of elements, using dynamically determined calibration factor from reference object in the image. Variation in perimeter of elements is used to perform classification of elements into target grain size, low size adulterants and higher size adulterants. The aspect ratio of target grain determines grain variety and adulterants determine adulteration percentage. Elements are classified into grain colored and non-grain colored adulterants. Grain colored adulterants are further classified as Grain Like Impurities and non-GLI, using predefined ranges of standard deviation of perimeter metric.

Abstract: The present invention relates to polysilazane compositions, in particular to a composition, the use of said composition for coating a substrate, and a substrate comprising a coating made from said composition.

Abstract: Image Retrieval is an application of computer vision that deals with searching images in large databases. Conventional methods utilize the entire image to perform the image retrieval task rather than considering specific features. The embodiments herein provide a method and system for feature based image retrieval. Initially, the system receives an input image and a query label. Further, a feature specific encoder is selected from a plurality of feature specific encoders based on the query label. A first set of feature vectors are computed from the input image using the selected feature specific encoder. Further, a Locality Sensitive Hashing (LSH) value is computed from the first set of feature vectors. Finally, a plurality of matching images is obtained from a plurality database images based on a comparison between the computed feature specific LSH value and a plurality of feature specific LSH values stored in a feature specific LSH database.

Abstract: A computer-implemented method, computer-readable medium, and computer system that involve operations including receiving, from a computing cluster, a first request for changing a first object type specified by an approval policy resource, where the approval policy resource is a first one of a plurality of custom resources; identifying, in response to receiving the first request and using a second custom resource of the plurality, an approval process associated with the first object type; performing the approval process associated with the first object type; and sending, to the computing cluster, a first notification indicating that the request has been approved.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods for synchronizing timestamps associated with media files across a variety of capture devices. For example, the disclosed systems receive a media file captured with a user device and determine, from metadata, initial timestamp information for the media file that includes inaccurate or incomplete time zone information. The disclosed systems determine a predicted time zone for the media file by extracting GPS information from the metadata associated with the media file or identifying a peer media file and utilizing a time zone or GPS information associated with the peer media file. The disclosed systems generate a synchronized timestamp for the media file with updated timestamp information based on the predicted time zone.

Abstract: Techniques and solutions are provided for increasing the energy efficiency of computing code. Many computing operations can be implemented in a number of different ways. While the end result of each implementation may be the same, the energy efficiency of the implementations can vary dramatically. Disclosed techniques provide rules that can be used to analyze code for a particular implementation of an operation. If a rule is triggered, a recommendation to replace the implementation with a more energy efficiency implementation can be provided, or the code can automatically be changed to include the more energy efficient implementation. Techniques for defining rules are also provided, such as by measuring or estimating energy used by various ways of implementing an operation. Measurement or estimation of energy used during code execution can be performed in a containerized environment, such as to provide improved accuracy, and allow for parallelization.

Abstract: A computer-implemented method for the numerical simulation of a heat exchanger with a numerical model comprising a plurality of elemental volumes such that they all correspond to the same configuration such that the joint of all the elemental volumes reproduces the volume in which heat transfer occurs between a hot fluid and a cold fluid. The simulation method establishes an iterative method on the set of elemental volumes in such a way that the relevant variables of each volume are updated using the data provided by the interpolation module without the need to run a classical simulation on the flow. The result is the assessment of the variables of interest in the entire volume of the heat exchanger for predetermined operating conditions.

Abstract: Examples described herein include systems and methods for contextually providing automated device enrollment into a management system. A management application on a user device can receive network settings for connecting to a local server. The network settings can be preconfigured by an administrator. The management application can cause the user device to send an enrollment request and a device identifier to the local server. The device identifier can be used to validate the device and provide a security token to the management application. The management application can use the security token to complete enrollment of the user device.

Abstract: A system identifies data objects from a first piece of code in an original programming language and creates a knowledge graph that represents a relationship among the data objects. The system identifies an expression syntax in the first piece of code and determines an interpretation of the expression syntax. The system creates a sentence that indicates an interpretation of a respective code line. The system determines a counterpart data object in a target programming language for each word in the created sentence. In response, the system creates a second piece of code in the target programming language, such that the second price of code satisfies syntax rules in the target programming language. In response to determining that the second piece of code is configured to perform a particular task that the first piece of code is configured to, the system executes the second piece of code.

Abstract: A system for processing patient data associated with a joint of a patient. The system comprising a computing device executing instructions to generate a 3D patient bone model of the joint in a non-weighted pose. The instructions rearrange 3D first and second bone models in order to mimic a weighted pose of the joint of the patient from a 2D image of the joint of the patient in a weighted pose by performing the steps of: generating a plurality of 2D projections of poses of the 3D patient bone model; comparing the plurality of 2D projections to contour lines of the first bone and the second bone in the 2D image; identifying particular 2D projections that best-fit a shape and size of the contour lines; and arranging the 3D first and second bone models relative to each other according to orientations represented by the particular 2D projections that were identified.

Abstract: The present relates to organic compound(s) [i e compound of formula I or compound of formula IA] and/or a composition comprising said organic compound(s) useful for utilization of non-coking coal in a blend comprising coking coal without deterioration of properties of coking coal, wherein said organic compound(s) is capable of releasing hydrogen in the plastic region of coal. The present disclosure provides a simple, economical, non-toxic and an efficient method for the preparation of said organic polymer and a method for utilization of non-coking coal in blend comprising coking coal without deterioration of properties of coke.

Abstract: An Application Data Management System (ADMS) enables an application file system to be mounted at any selected reconstruction time (TR). If the reconstruction time TR falls intermediate snapshot creation timepoints, the ADMS creates a version of the application file system at the selected reconstruction time TR using a snapshot of the data file from a previous application file system snapshot creation timepoint, and a snapshot of the log file from a subsequent application file system snapshot creation timepoint. The ADMS uses the snapshot of the log file from the subsequent snapshot creation timepoint to replay transactions on the snapshot of the data file from the previous snapshot creation timepoint up to the selected reconstruction time TR. This enables the state of the application file system to be recreated and mounted at any arbitrary selected reconstruction time, even if the selected reconstruction time is not coincident with snapshot creation timepoints.

Abstract: Historical information features that each comprise a supply chain performance indicator (such as a case fill rate performance indicator) for each of a plurality of different temporal windows are accessed and then at least some of the historical information features are weighted differently for at least some of the historical information features according to at least a first criterion to provide a training corpus. At least one machine learning model can then be trained using the training corpus to generate a machine learning model (or models) that are configured to predict supply chain performance.

Abstract: Techniques for combining the functionality of fabric interconnects and switches (e.g., Top-of-Rack (ToR) switches) into one network entity, thereby reducing the number of devices in a fabric and complexity of communications in the fabric. By collapsing FI and ToR switch functionality into one network entity, server traffic may be directly forwarded by the ToR switch and an entire tier is now eliminated from the topology hierarchy which may improve the control, data, and management plane. Further, this disclosure describes techniques for dynamically managing the number of gateway proxies running on one or more computer clusters based on a number of managed switch domains.

Abstract: The techniques disclosed herein enable systems to efficiently interface with zoned namespace (ZNS) storage devices through a specialized management entity. To achieve this, the management entity receives write requests from a file system containing data intended for storage at the ZNS device. In response, the management entity selects a zone from the ZNS device to write the file data to. Accordingly, the file data is written by appending the file data to the zone at a location indicated by a write pointer. When the write operation is completed, the offset of the file data within the zone is observed and recorded by the file system in file metadata. In contrast to typical systems which allocate locations at the storage device prior to writing, appending file data and then recording the location enables improved efficiency in file system operations. Namely, that write operations can be issued to the ZNS device non-serially.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods for utilizing machine learning models to generate modified digital images. In particular, in some embodiments, the disclosed systems generate image editing directions between textual identifiers of two visual features utilizing a language prediction machine learning model and a text encoder. In some embodiments, the disclosed systems generated an inversion of a digital image utilizing a regularized inversion model to guide forward diffusion of the digital image. In some embodiments, the disclosed systems utilize cross-attention guidance to preserve structural details of a source digital image when generating a modified digital image with a diffusion neural network.