Authors - Syamanarayanaprabha S S, Abymon A, Aarathy K Sudhakaran Abstract - Digital transactions in India have increased significantly since the National Payments Coperation of India introduced the Unified Payments Interface (UPI) in 2016. This study focuses on Generation X (ages 45-60) in Kerala, a state known for its high level of digital literacy, whereas adoption studies have mostly focused on younger generations like Millennials and Gen Z. The results, which are based on responses from 132 participants, show that security, transaction speed and convenience of use are important element that promote UPI adoption within this demographic. However, technological difficulties and security-related worries still serve as obstacles. To increase UPI acceptance in this group, the study emphasizes the need of more user-friendly application designs, focused digital education initiatives, and calculated interventions.
Authors - Salvatore Vella, Fatima Hussain, Salah Sharieh, Murrium Zaheer, Alex Ferworn Abstract - Artificial Intelligence (AI) using Large Language Models (LLMs) is now used to automate simple knowledge worker tasks such as creating communications and summarizing documents. This paper presents a domain-specific language for creating agents, which we call ContentCreator (CC), to perform common knowledge worker tasks and mimic the behaviour of knowledge workers and their workflows. An interpreter has been built to execute and automate these workflows. In the same way that computer languages have progressed from assembly language to increasingly higher-level languages, CC enables the high-level programming of workflows that utilize large language models. CC is a simple language that non-programmers can use and understand. We use several use cases to demonstrate the ability of ContentCreator to automate workflows across multiple tasks.
Authors - Bryan Atunca, Henry Coico, Edgar Ramos, Araceli Macassi, Donovan Fuqua Abstract - The adoption of digital technologies and intelligent systems is significantly enhancing efficiency and safety across the global agri-food industry. Integrating Industry 4.0 objectives with circular economy principles offers effective solutions to challenges such as resource scarcity and environmental impact. This study proposes a leagile supply chain model—combining lean and agile method-ologies—to optimize mango supply chains worldwide. By leveraging Cyber-Physical Systems (CPS) in postharvest processes such as drying and storage, the model enables remote monitoring and control of key environmental parameters. CPS technologies, through sensors, actuators, and data-driven algorithms, provide real-time feedback, minimize human error, and support precision in decentralized processing units. Findings show a reduction in total supply chain time from 41 to 30 days (26.8%), with efficiency gains of 25% in harvesting and 28.5% in storage. The integration of CPS strengthens circular supply chain practices, reduces losses, and enhances sustainability, profitability, and system resilience across global mango supply chains.
Authors - Abhijit T. Somnathe, Parameshwari M, Sampurnima Pattem, Sridhar N Koka, Shokhida Abdurakhmanova, Pundru Chandra Shaker Reddy Abstract - Gold prices are non-linear, unexpected, volatile, and unregulated, making prediction difficult. Numerous studies have anticipated gold prices because they affect international economic and monetary systems. However, linear relationship studies rarely explain gold price changes. Gold price time series data is unpredictable, nonlinear, and volatile, making prediction difficult. Classical statistics and machine-learning (ML) approaches like Random-Forests, CNNs, and RNNs offer excellent accuracy but limits. A model that combines Temporal-Convolutional-Networks(TCN) with Query (Q) and Keys (K) attention-mechanisms (TCN-QV) is presented to improve gold price forecasts. The model extracts temporal properties from sequence data using stacked dilated causal convolution layers in the TCN architecture. To adapt weight distribution to information features, an attention mechanism is introduced. Finally, dense layers give projected results. This approach predicts Shanghai gold price time-series. The optimised model reduces Mean Absolute Error (MAE) by 5.47% in the least favorable case and 33.69% in the most favorable case across four investigational datasets compared to the baseline model.
Authors - Pooja Anil Patil, Yashodhara Haribhakta Abstract - Named Entity Recognition (NER) is a core task in Natural Language Processing (NLP) with applications in conversational AI, healthcare analytics, and search systems. It enables structured extraction of key entities from text, supporting clinical decision-making and automated diagnosis. While NER resources for high-resource languages like English are widely available, low-resource languages such as Marathi lack domain-specific datasets, particularly in the mental health sector. This work introduces ArogyaMINDNER, a Marathi mental health-specific NER dataset, designed to enhance entity recognition for diseases, symptoms, and treatments. The dataset is manually annotated using BIO tagging, ensuring linguistic accuracy and contextual consistency across formal and conversational text. Annotation policies account for psychiatric terminology nuances, accommodating Marathi’s morphological richness and linguistic diversity. To assess performance, ArogyaMIND-NER is benchmarked using classical machine learning models (SVM, Naïve Bayes, XGBoost, LightGBM, Random Forest), deep learning architectures (CNN, BiLSTM, BiLSTM-CRF), and transformer-based models (mBERT, XLMRoBERTa, IndicBERT, MahaBERT, MahaMarathi-7B). Comparative evaluation reveals MahaMarathi-7B and Marathi-Social-NER as topperforming models, achieving higher precision and recall in disease recognition, while classical models struggle with contextual understanding. This dataset and trained models establish benchmarks for Marathi NLP research, contributing to mental health analytics, clinical applications, and low-resource language processing. The findings advocate for finetuning transformer architectures for domain-specific NER, ensuring better entity identification in Marathi mental health texts. The dataset and models will be made available for further research and development.
Authors - Navin Barde, Deepak S. Sharma Abstract - Good E-Learning system for colleges would ensure student learning to be flexible, accessible, and interactive for being remote learning option in the Post COVID-19 times. The most important aspects are an intuitive and easy to use GUI to easily browse on any device, the interactive content such as video lectures, simulations, quizzes, forum to improve the engagement. Customized learning paths, progress tracking, grade management, and instant communication features enhance learning. We take security very seriously, we encrypt all of your data while storing it and ensure that students privacy is maintained with strong user authentication. Laying particular emphasis on usability, interactivity, personalization, and security, such systems create a convincing and flexible learning environment, which encourages life long learning.
Authors - Mohit Menghnani Abstract - The increasing expenses and shortage of qualified developers have driven many to look to software development with an eye toward the incorporation of Artificial Intelligence (AI). Full-stack development facilitates the web application process, from the interface to the server-side functionality that powers it. AI has changed the tasks for many software developments services, and full-stack is one of them. The integration of Generative AI (GAI) into full-stack applications presents a transformative opportunity for healthcare assistance by enhancing conversational user interfaces. This study explores the development of an AI-powered healthcare assistant using a React.js frontend, a Node.js/Express.js backend, and Google GAI for symptom analysis and response generation. The aim is to assess the feasibility, performance, and effectiveness of Generative AI in providing real-time medical suggestions. A process of repeating development occurs and the system continuously improves through testing alongside user engagement. Users can maintain safe communication between systems through Jason Web Token (JWT) authentication and HyperText Transfer Protocol Secure (HTTPS) encryption features in addition to multi-language compatibility. Test results show that symptoms processed by the AI-enabled chatbot require 1.5 to 2.5 seconds before providing recommendations with 85-90% accuracy which surpasses traditional rule-based chatbots. The obtained results indicate how GAI shows promise to expand healthcare availability to wider populations.
Authors - Aneri Pandya, Killol Pandya, Hemant Yadav Abstract - Convolutional Neural Networks (CNNs) play a crucial role in medical image analysis, particularly in the detection of pneumonia, by facilitating accurate diagnosis and treatment planning. This paper presents a robust approach to pneumonia classification using the publicly available RSNA Pneumonia Detection Challenge dataset. The proposed method employs an ensemble-based CNN framework that integrates multiple pre-trained architectures, specifically MobileNetV2, InceptionV3, and ResNet152V2, through a combination of stacking techniques and soft voting. This ensemble strategy aims to leverage the individual strengths of each model for improved performance. All models are trained and evaluated exclusively on the RSNA dataset, and their effectiveness is assessed using AUC, accuracy, and F1-score metrics. The results demonstrate that the ensemble consistently outperforms individual models in terms of prediction stability and classification accuracy. This work highlights the potential of deep model aggregation techniques to enhance diagnostic reliability in clinical pneumonia screening systems.
Authors - Sana S Santhosh, Ashitha Mohan, Vishnu N Abstract - Artificial Intelligence (AI) has transformed the cosmetics market by providing tailored suggestions based on client interests and behaviors. This study investigates the correlation between AI-driven personalization and consumer trust, emphasizing critical elements such as the precision of AI recommendations, transparency, and the degree of customization. Research indicates that AI-driven customisation markedly improves client trust, therefore influencing consumer pleasure and brand loyalty. Trust development in AI-driven recommendations is contingent upon perceived reliability, security, and transparency. The review emphasizes that AI transparency techniques, including explainable AI (XAI), enhance customer confidence, whereas excessive personalization may provoke privacy issues. Moreover, the precision of AI recommendations is strongly linked to consumer happiness and intents to repurchase, enhancing perceived expertise and diminishing decision-making anxiety. Brands utilizing AI-generated cosmetic advice must equilibrate customisation, openness, and security to cultivate enduring consumer trust. Future research ought to investigate the ethical ramifications of AI personalization and analyze the evolution of confidence in AI-driven cosmetic brands over time. With the ongoing advancement of AI technology, it is imperative to optimize algorithms for precision and ethical concerns to maintain client confidence in the beauty sector.
Authors - Amarnath K R, Nandini Nayakudi, Lekha S Nair Abstract - Accurate and timely identification of insect species is a foundational requirement for precision agriculture, ecological monitoring, and integrated pest management. This paper presents a novel deep learning architecture for fine-grained acoustic classification of insect species using multichannel wingbeat recordings. Targeting critical agricultural use cases, the model is designed to differentiate between two morphologically and acoustically similar pest species, Halyomorpha halys and Nezara viridula, and a beneficial pollinator, Episyrphus balteatus. The proposed approach utilizes log-Mel and Per-Channel Energy Normalization (PCEN) spectrograms as input features and integrates a residual convolutional backbone with Squeeze and Excite blocks and Transformer inspired self-attention modules to enhance spectro-temporal representation learning. Evaluation on a curated subset of the InsectSound1000 dataset demonstrates a classification accuracy of 88.73% and a macro averaged F1 score of 88.78%. Class wise analysis reveals that the model effectively captures subtle differences in wingbeat harmonics, achieving an AUC of 1.00 and AP of 0.99 for the pollinator class. The results suggest strong discriminative power across ecologically distinct classes and highlight the model’s potential for deployment in automated, non-invasive insect monitoring systems for real-time agricultural decision-making.
