Wipro Limited

Sourav Mudi

Rishav Das

This disclosure relates to method and system for optimizing memory requirement for training an artificial neural network (ANN) model employed for natural language processing (NLP). In one embodiment, the method may include receiving a plurality of training parameters and a plurality of model parameters, selecting a set of model parameters from among the plurality of model parameters for training the ANN model based on a characteristic and an architecture of the ANN model, masking the set of model parameters in one or more layers of the ANN model based on a set of pre-defined rules to generate a set of masked model parameters, determining an amount of memory required for training the ANN model based on the set of masked model parameters, and providing the set of masked model parameters for training the ANN model when the amount of memory required is less than a determined threshold.

Method and system for optimizing memory requirement for training an artificial neural network model

The present invention discloses a method and a system for managing cache memory. The system comprising a processor is configured to receive datasets from one or more applications, segregate the received datasets into one or more data blocks, identify a checkpoint from previously created checkpoints stored in a virtual cache corresponding to the one or more data blocks, wherein the checkpoints are previously created based on frequency of repetition of each of the one or more data blocks and association between the each of the one or more data blocks, recall a sequence of previously stored data blocks from main memory based on the identified checkpoint, and send the sequence of previously stored data blocks to the one or more applications for execution, thereby managing cache memory.

System and method for managing cache memory

A method and system for automating repetitive task on a user interface is disclosed. The method includes identifying a plurality of Document Object Model (DOM) elements from a repetitive pattern. The method further includes identifying a set of dynamic DOM elements from the plurality of DOM elements. The method further includes determining a path and a path position within the application code of the repetitive pattern for each of the set of dynamic DOM elements. The method further includes training an Artificial Intelligence (AI) model to identify the repetitive pattern and to identify the corresponding path and the path position within the application code component, for each of the set of dynamic DOM elements. The method further includes automatically executing a code for each of the set of dynamic DOM elements based on the associated path and path position.

Method and system for automating repetitive task on user interface

A system and method for auto resolution of errors during compilation of data segments is disclosed. The method includes the steps of detecting one or more errors to be resolved associated with a data segment, determining an error type for each of the one or more errors from predefined error types based on neuron activations and a code logic derived from the data segment, arranging the one or more errors in a sequence of resolving, predicting a solution for each of the one or more errors, resolving the one or more errors sequentially for generating an error free data segment, and optimizing the error free data segment. The method further includes generating a pseudo code corresponding to the error free data segment and converting the pseudo code into at least one of the plurality of native languages.

System and method for auto resolution of errors during compilation of data segments

This disclosure relates to method and system for annotating tokens for natural language processing (NLP). In one embodiment, the method may include segmenting a plurality of corpus based on each of a plurality of instances, deriving a plurality of entities for each of the plurality of instances based on at least one of a machine learning technique or a deep learning technique, determining a word vector for each of the plurality of entities associated with each of the plurality of instances, and labelling a plurality of tokens for each of the plurality of instances. It should be noted that the plurality of tokens associated with the plurality of entities may be identified based on a frequency of each of the plurality of entities.

System and method for annotation of tokens for natural language processing

Disclosed herein is method and system for managing storage space complexity in a storage unit. In an embodiment, operational parameters related to memory operations and storage parameters related to memory blocks of the storage unit are analyzed to estimate storage capacity of each of the memory blocks. Subsequently, the memory blocks are clustered into plurality of clusters based on the storage capacity. Further, one or more of the plurality of clusters are selected for performing future memory operations based on ranking of the plurality of clusters. In some embodiments, the present disclosure helps in dynamically managing storage space complexity in the storage unit and optimizes the storage space utilization. Also, the present disclosure automatically handles storage volumes, thereby reducing latency in memory backup operations and reducing amount of buffer/cache memory required.

Method and system for optimizing storage space in a storage unit

Disclosed herein is a method and Serially Attached SCSI (SAS) controller for transmitting data using SCSI. In an embodiment, a plurality of I/O operations received from a storage unit are fragmented into a plurality of blocks. Further, each of the plurality of blocks are mapped with corresponding memory drives. Thereafter, a reduced number of virtual lanes required for transmitting the plurality of blocks to the corresponding memory drives is estimated. Finally, the reduced number of virtual lanes are created for transmitting the plurality of blocks to the corresponding memory drives. In an embodiment, the present disclosure uses virtual lanes for transmitting data, thereby eliminating requirement of dedicated, physical lanes for transmitting the data. Consequently, according to embodiments of present disclosure, the SAS controller may be configured to simultaneously activate multiple virtual lanes for completing the data transmission, thereby resulting in faster and reliable data transmission.

Method and system for transmitting data using small computer system interface (SCSI)

A method and system of transmitting packets within a network is disclosed. The method includes determining frequency and sequence of transmission for a plurality of packets within the network. The method further includes identifying a set of recurrent packets from the plurality of packets based on the determined frequency and sequence of transmission. The method includes training a neural network based on the identified set of recurrent packets to identify recurrent packets. The method further includes generating a packet checkpoint based on the trained neural network. The method includes regenerating one or more of the set of recurrent packets in response to a request for the set of recurrent packets, based on the packet checkpoint.

Method and system for transmitting packets within network

A method of managing Input Output per Second (IOPS) in Fiber Channel network topology is disclosed. The method includes initializing a plurality of network parameters into dependent variables and independent variables. The method includes performing multilinear regression analysis on at least one of the dependent variables and the independent variables to predict a value of at least one dependent variable. The method includes clustering each of a plurality of switches into a plurality of categories based on the predicted value. The method includes determining a plurality of routes between source and destination switches associated with an lops request. The method includes establishing a connection between the source and destination switches through a shortest route including a subset of switches. The method includes modulating, for each switch in the subset, a bandwidth according to an IOPS associated with the IOPS request and a category associated with each switch in the subset.

Method and system for managing input output per second (IOPS) in fiber channel network topology

The present subject matter relates generally to optimization of distribution of data packets in a High Performance Computing (HPC) switch. The HPC switch receives data packets from controllers communicatively connected to HPC switch, through interconnected Queue Pair (QP) lanes and assigns a destination virtual lane (VL) for each data packet based on service level information. Further, the HPC switch generates plurality of clusters of ports that are coupled with each interconnected VL based on latency determined dynamically for each port. Finally, the HPC switch transmits a first group of data packets through ports belonging to a primary cluster of plurality of clusters and transmits acknowledgement key-packet associated with second group of data packets through the ports belonging to secondary clusters of plurality of clusters, based on destination VL assigned for each data packet to other controllers for optimizing the distribution of the plurality of data packets in the HPC switch.

Method and high performance computing (HPC) switch for optimizing distribution of data packets

A method and system for recovering data from a storage system is disclosed. The method includes predicting, by a fault tolerance device, a fault within the storage system comprising one or more disk array enclosures, wherein each of the one or more disk array enclosures comprise a plurality of disks. The method further includes identifying, by the fault tolerance device, a data block going to be affected by occurrence of the predicted fault, wherein the data block is identified within a first disk in the plurality of disks. The method includes extracting, by the fault tolerance device, data stored in the data block before occurrence of the predicted fault. The method further includes transferring, by the fault tolerance device, the data extracted from the data block to a second disk not going to be affected by occurrence of the predicted fault.

Growing community of inventors

Burdwan, India

Sourav Mudi