Tata Consultancy Services

Conventional techniques for evaluating generated infographic images, prioritize aspects like text alignment, data adherence, aesthetics, and overall quality, but often do not comprehensively assess adherence to underlying input data. Embodiments herein provide a method and system for comprehensive assessment of infographic images using Aesthetics Adherence to Data and Text (AADaT) Score. The method calculates a plurality of AADaT scores using the plurality of attribute scores for the plurality of generated infographic images. Further a plurality of LLM scores along with a plurality of feedback instructions are generated for each of the plurality of attributes, using the associated plurality of AADaT scores for the plurality of infographic images. An infographic evaluation model is trained with the plurality of infographic images, the plurality of feedback instructions, the plurality of AADaT scores, and the plurality of LLM scores, to generate a trained infographic evaluation model.

Existing design methods rely on manual effort, trial-and-error approaches, and simplified models, which can lead to suboptimal designs and reduced efficiency. The present disclosure obtains a first set of responses for a first set of questions from a user and generates a second set of questions. A second set of responses is obtained and one or more questions specific to a problem formulation are generated using one or more autonomous programs. A first set of data corresponding to one or more questions is retrieved from a knowledge base and one or more gaps is identified. A second set of data is retrieved from one or more resources and knowledge base is updated. One or more process models and corresponding one or more optimization models are retrieved, and a relevant process model is selected. The most appropriate one or more optimization models are selected and, and feedback is obtained from user.

The growing use of wearable devices requires accurate and compact representations of high dimensional physiological signals. The present disclosure presents a UNet inspired autoencoder to represent and reconstruct multiple neurophysiological signals from single channel data. The architecture includes single-encoder and multiple decoders to obtain self-attention enabled compact embeddings of mixed ExG (EEG /ECG) signals through decaying encoder-decoder skip connections, for improved representation capability. The embeddings are separable into individual ExG components enabling simultaneous reconstruction of high-fidelity EEG and ECG sources. The pretrained encoder can be used for a complex downstream task with minimum fine-tuning. Using the present disclosure on a large corpus of single-channel mixed ExG generated from overnight Polysomnography (PSG) recordings, it is shown that subject- and class- independent EEG/ECG reconstructions validated by multiple domain-specific metrics can be obtained, and the encoded EEG embeddings can be classified into five sleep stages as a downstream task.

This disclosure relates generally to screening large data-sets with multi-level clustering. The State-of-the-art methods leveraging LLMs with generative capabilities focus on scanning an extraneous and peripheral contents in text documents to retrieve contextually relevant information. This leads to hallucinations and uncertainty in response and results in complicating the generation of actionable insights. The method of the present disclosure performs a two-level clustering at both an intra-document and an inter-document levels to extract context relevant data. The method screens semantically similar sentence chunks at intra-document level to create a concise intra-document embeddings with contextually relevant text data. Further, an inter-document level clustering is performed by computing a cluster integrity and relevancy (CLIAR) index on a plurality of intra-document embeddings to isolate inter-document cluster having semantic representation of contextually relevant contents.

The embodiments of the present disclosure address unresolved problems of precise and scalable anomaly detection, root cause analysis, anomaly severity estimation, and trend analysis for plants within controlled environments. Embodiments herein provide a method and system for estimating severity and causality for plant development anomalies in plants in a controlled environment. The system uses RGB image based analysis to identify subtle, similar looking growth anomalies and timely corrective actions. System estimates anomaly severity by analyzing real-time data and dynamically generated synthetic baseline plant data for non-linear severity growth for plant anomalies. The system performs root cause analysis with prioritized rankings, severity zone mapping, severity trend estimation, and tailored role-specific insights to enable effective decision-making and targeted interventions in plant farming environments. Finally, the system identifies and ranks potential root causes of plant-specific stress symptoms along with providing a final zone, bench and rack health index and overall health status.

This disclosure relates generally to a method and system for predicting animal emotions using animal emotion knowledge graph and graph neural network. Current available methods focus only on visual and language data captured from animals and lacks real time adaptability. The method disclosed generates an animal emotion knowledge graph (AEKG) that combines human and animal neurobiological data, behavioral studies, and the human wheel of emotions. Further real time graphs are generated from multimodal input data captured from the animal. These real time graphs are used for predicting primary, secondary, and tertiary emotions of the animals using a trained graph neural network-Transformer model. This model is trained using the AEKG. Using temporal graph analysis, the method predicts future emotions and generates real-time recommendations based on generative artificial intelligence techniques. Predicting the emotions of animals in real time helps to grasp their emotional well-being to improve their care and management effectively.

This disclosure relates generally to a method and system for conducting an offline token-based digital currency payment. State-of-the-art methods facilitating offline transaction lacks validity of each token involved in the transaction. Moreover, a separate record is required for the transactions processed in an online mode and for the transactions processed in an offline mode. Further, managing multi-hop offline transactions are not yet achieved. The disclosed method involves offline transfer of set of tokens from a payer's wallet to a payee's wallet. The method captures one or more transaction dependencies associated with each token and assigns a token-proof to each token in the set of tokens. The method executes the transaction having cleared or no prior transaction dependency by transmitting the set of tokens from the payer's wallet to the payee's wallet using a payer's WSP, a central authority, and a payee's WSP.

Conventionally, to automate these laboratory tasks, a robotic device has been used which is not enough as the sequence of actions like pipetting and dipping are more complex than just having grasping capabilities. Present disclosure provides a liquid handling system (LHS) and a method that determines quantity for formulation to be prepared using liquids associated with recipe. An error between a set point and actual volume of the liquids is calculated for determined motor speed to generate torque and rotational motion that are used to generate a linear motion for movement of one or more plungers associated with one or more syringes mounted to the LHS. The movement of plungers enables aspiration and dispensation of the liquids by the LHS at a required amount for preparing the quantity of the formulation using the viscosity of the liquids.

Existing approaches for determining user interest in advertisements being displayed have the disadvantages related to complexity, scalability, and so on. Method and system disclosed herein obtain a video of a user, and further processes to generate an attention series data representing a plurality of visual attention values of the subject, based on an Eye Aspect Ratio (EAR) and a Blink Rate Variability (BRV) score. Further, a fluctuation matrix comprising a plurality of attention states of the subject is generated. Further, an attention stability frequency score representing frequency of oscillations in the attention series data is generated, and then an attention stability power score representing an intensity of occurrence of attention values in the attention series data is generated. The attention stability frequency score and the attention stability power score are fused to generate a like-dislike score indicating an interest of the subject in the advertisement being played.

Existing microservices identification techniques focused on clustering the cohesive components of the applications, ignoring their alignment to the business domain that they service. The present disclosure analyzes a plurality of inputs and generates a plurality of outputs based on the analysis. A plurality of application elements and one or more associated relations of different types between the plurality of application elements is created using the generated plurality of outputs. A filtered set of application elements and the associated relations of different types between the set of application elements are enriched. An application architecture model is created using an enriched set of application elements and an architecture meta model. A weighted graph is created using the application architecture model and a domain description map. One or more communities specific to the created weighted graph are grouped and evaluated. One or more candidate microservices are identified based on the evaluation.