Tata Consultancy Services Patente

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

To maintain a healthy and safe battery, the Battery Management System (BMS) require numerous sensors which is a costly process. Immersion cooling applications where the battery stack is cooled with a dielectric coolant fluid has complex flow patterns that are not modeled easily without high fidelity modeling. The present disclosure provides a solution by combining physics-based and data-driven techniques. Initially, a first reduced subspace and a second first reduced subspace are extracted from the high-fidelity or full-order model (FOM) data with the help of proper orthogonal decomposition (POD) in offline. Further, the dynamics of the reduced spaces (first reduced subspace and the second reduced subspace) are captured by computing the reduced coefficients. The reduced coefficients are computed by solving the reduced order equations. The reduced model is solved in the same manner as the high-fidelity or FOM model. The reduced-order solution is constructed once the reduced coefficients are obtained.

Multi-object tracking (MOT) in video sequences plays a critical role in various computer vision applications. The primary objective of MOT is to accurately localize and track objects across consecutive frames. However, existing MOT approaches often suffer from computational limitations and low frame rates in commodity machines, which hinders real-time performance. Present disclosure provides method and system for performing content aware multi-object tracking. The system first classifies video into slow and fast moving object content videos depending on features of objects to be tracked in frames. Then, system applies a computationally intensive deep sort algorithm to perform tracking of objects by selectively skipping frames. Thereafter, the system applies linear approximate Kalman prediction for slow object content videos and quadratic interpolation for fast object content videos as low computation tracking techniques for tracking objects present in skipped frames, thus significantly improving execution speed while reducing computational load on the system.

Realistic data is an important aspect for training vision based systems. Conventional approaches need complex prerequisites to generate such data and are quite expensive as well. The present disclosure provides one stop solution for different artificial intelligence rendering, scenario generation and future video prediction. The present disclosure provides a provision for fusion of Generative Artificial Intelligence (GenAI), deep learning and image processing techniques and explores the possibilities of generating data from simulated, real, or fused environments. Further, the present disclosure utilizes a panoptic Segmentation approach to create semantic labels and a flow map-based methodology taking cues from a sequence of frames helps to address long time temporal coherence which is a key issue in generated data. This goes as an input to the Generative AI network which generates synthetic dataset based on the aspects of dynamic objects, scene changing and environment variations.

The disclosure generally relates to methods and systems for graph assisted unsupervised domain adaptation for machine fault diagnosis. The present disclosure solves the technical problems in the art using a Graph Assisted Unsupervised Domain Adaptation (GA-UDA) technique for the machine fault diagnosis. The GA-UDA technique carries out the domain adaptation in two stages. In the first stage, a Class-wise maximum mean discrepancy (CMMD) loss is minimized to transform the data from both source and target domains to a shared feature space. In the second stage, the augmented transformed (projected) data from both the source and the target domains are utilized to construct a joint graph. Subsequently, the labels of target domain data are estimated through label propagation over the joint graph. The GA-UDA technique of the present disclosure helps in addressing significant distribution shift between the two domains.

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

The present disclosure herein provides methods and systems for optimizing personalized hospitality offerings to suit based on the customer requirement. The present disclosure employs a bucket of prediction models, namely (i) pre-trained hotel prediction model for predicting one or more hotels present in a destination city, (ii) the pre-trained room prediction model for predicting the one or more vacant rooms from the one or more hotels, and (iii) the pre-trained ancillary services prediction model for the predicting the one or more ancillary services available for the one or more vacant rooms. Each prediction model is separately trained on the features obtained from the unstructured historical training data, using a feature extraction technique. The fluidic pricing mechanism is used to provide personalized hospitality offerings by determining the fluidic pricing and offers to multiple relevant ancillary service bundles which may suit mostly to the diverse customers.

State of the art deep network based Universal Lesion Detection (ULD) techniques inherently depend on large number of datasets for training the systems. Moreover, these systems are specifically trained for lesion detection in organs of a Region of interest (RoI) of a body. Thus, requires retraining when the RoI varies. Embodiments herein disclose a method and system for domain knowledge augmented multi-head attention based robust universal lesion detection. The method utilizes minimal number of Computer Tomography (CT) scan slices to extract maximum information using organ agnostic HU windows and a convolution augmented attention module for a computationally efficient ULD with enhanced prediction performance.

This disclosure relates generally to system and method for molecular property prediction. The conventional methods for molecular property prediction suffer from inherent limitation to effectively encapsulate the characteristics of the molecular graph. Moreover, the known methods are computationally intensive, thereby leading to non-performance in real-time scenarios. The disclosed method overcomes the limitations of typical dynamic neighborhood aggregation (DNA) method by fusing the static edge attributes in determining the self-attention coefficients. In an embodiment, the disclosed method transforms the hidden state of a sink node by utilizing a neural-net function, which takes as input an aggregated single-message vector obtained by the self-attention mechanism and the self-attention mechanism transformed hidden state of the node.

This disclosure relates generally to system and method for molecular property prediction. The conventional methods for molecular property prediction suffer from inherent limitation to effectively encapsulate the characteristics of the molecular graph. Moreover, the known methods are computationally intensive, thereby leading to non-performance in real-time scenarios. The disclosed method includes performing self-attention on the nodes of a molecular graph of different sized neighborhood, and further performing a shared attention mechanism across the nodes of the molecular graphs to compute attention coefficients using an Edge-conditioned graph attention neural network (EC-GAT). The EC-GAT effectively utilizes the edge characteristics in the molecular graph for molecular property prediction.

Good health is one of the paramount essentials for any human being. A method and system for assessing and improving wellness of a person using body gestures have been provided. The system is configured to address fitness, wellness and physical movement related problems for both the abled and the differently abled society. The disclosure provides an intelligent and adaptive system that will comprehend the gestures and postures of the person to assess his capability of undertaking wellness activities. The person's movement capabilities and the environment are learnt by the system to track the gestures for the activities. Based on his gestures, the extent of activity achievement is calculated. The system also has the feature of provisioning dynamic activities to the person based on prior attempts and levels of achievements. Additionally, the system also providing the fitness routine in an AR environment by human-to-object mapping through object detection and augmentation.

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

This disclosure relates generally to safety improvement in a workspace. The method includes creating a context model associated with the workspace by defining contextual domain and a plurality of safety violation rules for safety standard operating procedures (SOPs) of digitally augmented assets corresponding to the physical entities present in the workspace. Further, the method includes monitoring, in real-time, states of each of the digitally augmented assets by obtaining sensor-based data from a sensor devices deployed in a smart space environment implemented in the workspace. SOP violations are triggered as events on a comparison of the states of digitally augmented assets with the safety violation rules using the context model.

Data inaccuracy and insufficiency are critical aspects to be analyzed to improve batch predictions, specifically in context of SLA jobs as they are foremost in affecting deliverables. Embodiments of the present disclosure provide a method and system for enhancing batch predictions by localizing jobs contributing to time deviation and generating fix recommendations by fixing data inaccuracy and insufficiency. The term fix recommendation refers to recommending a list of plausible fixes to identified causes that reduce batch prediction errors enhancing accuracy of predictions. The localization is performed by bottom-up traversing of a batch graph representing a batch process, if the batch process has a Service level Agreement (SLA) job, by narrowing down to the SLA job that has end time inaccuracies. The localization enables identifying the origin or real contributors and root cause analysis is performed for the localized jobs to generate effective fix recommendations by fixing data inaccuracy and insufficiency.

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

Zusammenfassung wird geladen …

The present disclosure estimates items to be returned based on nature of items picked online in association with the context. The context includes time, shopper details, local events and the like. It is addressed by mapping return spread and similarity spread or linkage spread in unique format. Further, the intrinsic mechanism that result returns are captured by training a multivariate Machine Learning (ML) model using the actual return spread, the similarity spread, or the linkage spread and the customer profile data. The captured return mechanism is leveraged to pre-empt the returns online in the form of return spread at the time of ordering in real time.

Zusammenfassung wird geladen …

High-performance deployment of DNN recommendation models heavily rely on embedding tables, and their performance bottleneck lies in the latency of embedding access. To optimize the deployment of RMs, the method and system is disclosed, which leverages heterogeneous memory types on FPGAs to improve the overall performance by maximizing the availability of frequently accessed data in faster memory. The system, using a optimizer dynamically allocates table partitions of the embedding tables based on history of input access history. A pre-optimizer block disclosed determines whether smaller tables should be partitioned or placed entirely in smaller memories, improving overall efficiency. The performance of RM is improved with improvement in average embedding fetch latency and effectively inference latency via modified Round Trip computation.

This disclosure relates generally to neural network inferencing, and more particularly, to a method and system for neural network inferencing in logarithmic domain. The conventional techniques include training a neural network in logarithmic domain and performing inferencing. This leads to less accuracy, challenge in converting large models and unable to perform optimization. The present disclosure converts a pre-trained neural network into logarithmic domain using a bit manipulation based logarithm number system technique wherein the neural network is pre-trained in real time or in logarithmic domain. The method converts the weights, neural network layers and activation function into logarithmic domain. The method uses a 32-bit integer variable to store a logarithm number which leads to memory efficiency. The disclosed method is used for inferencing of convolutional neural network for natural language processing, image recognition and so on.

Zusammenfassung wird geladen …

Use of ultrasonic guided waves for damage identification and localization is not new in Non-Destructive Testing/Evaluation. However, most of the time it is performed with high voltage pulse excitations that use several hundreds of volts in the form of a short burst, thus making it unsafe and unsustainable for defect localization in large structures. Present disclosure provides a method and a system for damage localization using low power ultrasonic guided waves. The system of the present disclosure uses a Vector network analyzer (VNA) sweep of a defined frequency range of low signal amplitude on a structure to form guided wave resonance spectra. Then, the system performs an Inverse Fast Fourier transform (IFFT) on the guided wave resonance spectra to obtain a time domain pulse propagation picture. Thereafter, the system uses a pulse echo based analysis technique based on time domain pulse propagation picture to locate damage position in the structure.

Zusammenfassung wird geladen …

The embodiments of the present disclosure herein address unresolved problem of obtaining price recommendation not only for the immediate moment but also for a continuous period until a market-driven price trigger necessitates change. Embodiments herein provide a framework for dynamic pricing of one or more products in an e-commerce retail using at least one contextual Bandit (CB) technique. The framework provides an ensemble of contextual bandits for dynamic pricing in e-commerce. The framework amalgamates strengths of one or more CB models. The framework incorporates an ensemble layer to critically review immediate pricing suggestions, ensuring their long-term applicability, thereby balancing margin stability and customer satisfaction over time.

Zusammenfassung wird geladen …