Abstract: A remote data recovery system is determined to be unsuitable for communications. A stream of source system storage changes associated with an object is received at a backup system from a source system while the remote data recovery system is unsuitable for communications. The backup system is utilized to generate one or more reference restoration points based on the stream of source system storage changes associated with the object. The remote data recovery system is determined to be suitable for communications. In response to determining that the remote data recovery system is suitable for communications, a hot standby of the object hosted by the remote data recovery system is updated to a reference restoration point generated by the backup system prior to the remote data recovery system becoming suitable for communications.

Abstract: Embodiments described herein provide a cross-lingual intent classification model that predicts in multiple languages without the need of training data in all the multiple languages. For example, data requirement for training can be reduced to just one utterance per intent label. Specifically, when an utterance is fed to the intent classification model, the model checks whether the utterance is similar to any of the example utterances provided for each intent. If any such utterance(s) are found, the model returns the specified intent, otherwise, it returns out of domain (OOD).

Abstract: The present disclosure describes a system and method of developing a search database for automated content curation. The processing arrangement is configured to store a plurality of slides related to one or more fields in the database arrangement, generate a plurality of thumbnails from the stored plurality of slides, extract a plurality of metadata from the plurality of thumbnails, wherein the plurality of metadata are extracted by processing the generated plurality of thumbnails through an optical character reader, store the extracted plurality of metadata into the search database, create an index comprising of associations between the plurality of slides, the plurality of associated metadata and the generated plurality of thumbnails, store the created index in the search database and curate content from the index of the search database based on the one or more search strings.

Abstract: This disclosure relates generally to program synthesis for weakly-supervised multimodal question answering using filtered iterative back-translation (FIBT). Existing approaches for chart question answering mainly address structural, visual, relational, or simple data retrieval queries with fixed-vocabulary answers. The present disclosure implements a two-stage approach where, in first stage, a computer vision pipeline is employed to extract data from chart images and store in a generic schema. In second stage, SQL programs for Natural Language (NL) queries are generated in dataset by using FIBT. To adapt forward and backward models to required NL queries, a Probabilistic Context-Free Grammar is defined, whose probabilities are set to be inversely proportional to SQL programs in training data and sample programs from it.

Abstract: Methods, systems, and devices for data management are described. A data management system (DMS) may receive a request to access metadata stored in a distributed metadata repository. The request may have a first semantic format compatible with a first version of the distributed metadata repository. The DMS may translate the first semantic format of the request to a second semantic format compatible with a second version of the distributed metadata repository. The DMS may perform the semantic translation according to a label in the request. Accordingly, the DMS may determine whether a current version of the metadata corresponds to the first version of the distributed metadata repository or the second version of the distributed metadata repository, and may access the metadata in the distributed metadata repository based on the request and the current version of the metadata.

Abstract: A method for ensuring precedence for the processing of a blockchain transaction to prevent loss of cryptographic currency includes receiving a new blockchain transaction by a blockchain node in a blockchain network, confirming the new blockchain transaction including identifying a precedence transaction that was previously conducted and stored in the blockchain that involves both blockchain wallets included in the new blockchain transaction, including the new blockchain transaction in a new block that is generated, and distributing the new block to additional nodes in the blockchain network for confirmation and addition to the blockchain.

Abstract: A recovery orchestration pipeline has stages configured to control efficient failover and recovery of applications in a multi-site disaster recovery (DR) environment. The applications run on user virtual machines (UVMs) at a primary site of the DR environment and connect to block storage devices (BSDs) that export virtual disks over a storage protocol to consume data including a recovery plan for disaster recovery. The recovery plan includes a recovery configuration whose generation is triggered by a user via a graphical user interface (GUI) and specifies resource requirements needed to recover the applications at a secondary site in the event of a disaster. The orchestration pipeline is initiated via single click of the GUI and completion of the stages of the pipeline is displayed as progress via the GUI to allow recovery of the applications without user intervention.

Abstract: The need for extracting information trapped in unstructured document images is becoming more acute. A major hurdle to this objective is that these images often contain information in the form of tables and extracting data from tabular sub-images presents a unique set of challenges. Embodiments of the present disclosure provide systems and methods that implement a deep learning network for both table detection and structure recognition, wherein interdependence between table detection and table structure recognition are exploited to segment out the table and column regions. This is followed by semantic rule-based row extraction from the identified tabular sub-regions.

Abstract: Traditional systems that enable extracting information from Piping and Instrumentation Diagrams (P&IDs) lack accuracy due to existing noise in the images or require a significant volume of annotated symbols for training if deep learning models that provide good accuracy are utilized. Conventional few-shot/one-shot learning approaches require a significant number of training tasks for meta-training prior. The present disclosure provides a method and system that utilizes the one-shot learning approach that enables symbol recognition using a single instance per symbol class which is represented as a graph with points (pixels) sampled along the boundaries of different symbols present in the P&ID and subsequently, utilizes a Graph Convolutional Neural Network (GCNN) or a GCNN appended to a Convolutional Neural Network (CNN) for symbol classification. Accordingly, given a clean symbol image for each symbol class, all instances of the symbol class may be recognized from noisy and crowded P&IDs.

Abstract: A solenoid based air ventilation valve (24) comprises of a housing (1), a solenoid coil (2), a flexible diaphragm (6), a plunger (8), a filter inlet (28) and a filter outlet assembly (100). The flexible diaphragm (6) is over molded with the plunger (8) to seal the flexible diaphragm (6) against the high pressure during closed position and has an incorporated O-ring (102) with flexible diaphragm (6) to stop the leakage from the valve (24). The filter outer assembly (100) is fixed to the inlet (25) and outlet port (26) to prevent suspended contamination particles in the valve (24). The valve (24) has a press fitted metal insert/flow controller (55) to achieve a low flow rate. Further, the air ventilation valve (24) allows a low leakage limit, low flow rate to serve high opening and working pressure to work in both over pressure and under pressure (vacuum) conditions.

Abstract: The need for extracting information trapped in unstructured document images is becoming more acute. A major hurdle to this objective is that these images often contain information in the form of tables and extracting data from tabular sub-images presents a unique set of challenges. Embodiments of the present disclosure provide systems and methods that implement a deep learning network for both table detection and structure recognition, wherein interdependence between table detection and table structure recognition are exploited to segment out the table and column regions. This is followed by semantic rule-based row extraction from the identified tabular sub-regions.