Browse Items (7684 total)
Sort by:
-
Cognitive Load Optimization in Digital (ESL) Learning: A Hybrid BERT and FNN Approach for Adaptive Content Personalization
Traditional English as a Secondary Language (ESL) learning platform rely on static content delivery, often failing to adapt to individual learners cognitive capacities, leading to inefficient comprehension and increased cognitive load. A novel hybrid Feedforward Neural Network and Bidirectional Encoder Representation Transformer (FNN-BERT) framework stands as our solution because it performs dynamic content personalization through predictions of real-time cognitive load. The proposed approach incorporates Feedforward Neural Networks (FNN) alongside Bidirectional Encoder Representations from Transformers (BERT) to process behavioral analytics for optimized content complexity adjustment and adaptive and scalable learning delivery. Real-time adaptability, scalability and high computational needs of current models reduce their effectiveness in personalized learning environments. Through the application of Test of English for International Communication (TOEIC), International English Language Testing System (IELTS) and Test of English as a Foreign Language (TOEFL) datasets, our methodology uses Feedforward Neural Network (FNN) to forecast cognitive load based on student engagement behaviors and application errors then Bidirectional Encoders Representations from Transformer (BERT) processes content difficulty adjustments automatically. The proposed model delivers a 95.3% accuracy rate, 96.22% precision level, 96.1% recall capability and 97.2% F1-score which surpasses conventional Artificial Intelligence-based English as a Secondary Language (ESL) learning systems. The system makes use of Python for its implementation to improve understanding as well as student focus and mental processing speed. Personalized content presentation methods lead to lower cognitive strain which simultaneously advances student achievement numbers. The research adds value to smart educational frameworks through its introduction of a scalable framework that allows adaptable learning systems for English as a second language (ESL). The following research steps include simplifying system complexity while adding multimodal learning signals including eye monitoring and speech recognition and further developing the model across various educational subject areas. The research works as a promising foundation which propels AI real-time adaptive education systems for students from various backgrounds. (2025), (Science and Information Organization). All Rights Reserved. -
Digital Adoption and Price Discovery in Shadows: Evidence from Indian IPO Grey Markets between 2016-2025
This study examines the role of digital adoption and price discovery through informal IPO markets in India. Due to price anomalies, sentiments around the IPO listing day are channelised through Informal markets known as grey markets. This study verifies the determinants of grey market IPO prices and their linkages with the formal market IPO under-pricing. Apart from dominant market and firm-specific factors, it examines how digital adoption variables such as digital payment usage and new demat accounts affect the IPO prices in both channels. In the post-pandemic period, there is a surge in the number of IPOs offered and participation of institutions and individuals in IPOs. Grey market allows traders to bid on IPO applications before they are officially listed, helping to assess under-pricing in issue prices, if any. This study uses data of 1,155 IPOs that went public in India between 2016 and 2025. Using OLS models, the study examines the relationships among variables, and findings indicate that both grey market activity and digital adoption directly influence listing day pri ces, confirming that under-pricing is predictable through informal channels. Investors should consider these factors, in addition to fundamental aspects, when making IPO investment decisions. 2025, ASERS Publishing House. All rights reserved. -
Non-Accounting Drivers of Forensic Accounting Techniques: Insights from PLS-SEM Analysis
Forensic accounting techniques are pivotal in combating financial fraud and enhancing corporate governance. According to Forensic Accounting Theory, both accounting and non-accounting factors influence the intention to adopt these techniques. This study explores the impact of key non-accounting factors i.e. Bonus Contract, Anonymity, and Collapse Avoidance on adoption of forensic accounting techniques by the practitioners, employing Partial Least Squares Structural Equation Modeling (PLS-SEM) and SmartPLS software. Data was collected from professionals across diverse industries utilising forensic accounting services. The results reveal that these non-accounting factors exert varying levels of influence on adoption intentions. This research enriches the existing body of knowledge by offering new perspectives on the role of non-accounting drivers in forensic accounting adoption, providing actionable insights for policy-makers, regulators, and corporate leaders. 2025 The Author(s). -
From Fabric to Function: Decoding Athleisure Wear Adoption through Fit-inspired Lifestyle and Social Influence; [Od tkanine do funkcionalnosti: analiza sprejemanja portno-elegantnih obla?il skozi zdrav ivljenjski slog in drubeni vpliv]
Athleisure wear has emerged as a prominent trend in the fashion industry, particularly among educated youth in India, and is driven by a growing focus on fitness, health and comfort. This study, based on the stimulus-organism-response model, examines the factors influencing athleisure wear purchase intention, with a focus on fit-inspired lifestyle, perceived product quality, social influence and brand reputation. The study examines how these factors influence enclothed cognition, ultimately impacting purchase intention. Partial least squares-structural equation modelling, based on the two-step approach, was employed using SmartPLS 4.0 software for data analysis. Data collected from young consumers in Bangalore indicated that a fit-inspired lifestyle, perceived product attributes and social influence had a positive influence on enclothed cognition, while brand reputation did not show a significant relationship. Additionally, enclothed cognition was found to be a significant predictor of purchase intention. The results suggest that the alignment of athleisure wear with a fit-inspired lifestyle, functional attributes and social validation plays a crucial role in shaping purchase intention. The study offers practical insights for marketers to focus on lifestyle alignment, product functionality and social engagement in their marketing strategies. The findings also highlight a shift in consumer behaviour where experiential and value-driven factors, such as product benefits and lifestyle fit, outweigh traditional brand reputation. Future research should include exploring diverse demographic segments, longterm behavioural patterns and the impact of sustainability and cultural values on purchase behaviour in the area of athleisure wear. 2026, University of Ljubljana Press. All rights reserved. -
Enhanced Pneumonia Detection from Chest X-rays Using Machine Learning and Deep Neural Architectures
Pneumonia is a major worldwide health concern, particularly for vulnerable groups such as babies and the elderly. Despite advances in medical imaging, diagnosing pneumonia using a chest X-ray remains difficult, due to the subtle presentation of symptoms and the variety in picture interpretation. This study utilizes modern machine learning can improve the accuracy and speed of diagnosing pneumonia using chest X-ray images. Utilizing a comprehensive dataset from the Kaggle online repository, consisting of over 5,000 annotated images, we evaluate the efficacy of various machine learning models including deep convolutional neural networks (CNN) and ensemble learning techniques. Our findings indicate that models like the Fuzzy opponent histogram filter combined with Logistic model trees (LMT) achieved the highest accuracy at 96.97%, while the deep learning-based Lenet (CNN) with LMT closely followed at 95.85%. The study aims to improve diagnostic precision, reduce interpretation discrepancies, and facilitate faster clinical decision-making by identifying the most effective machine learning approaches for real-world applications in healthcare settings. 2025 Kamal Upreti, Anju Singh, Divakar Singh, Preety Shoran, Uma Shankar, Meenakshi Yadav and Rituraj Jain. -
Enhancing Image Classification Performance through Hybrid Self-Supervised Learning Strategies
Image classification is a cornerstone of computer vision, with the applications spanning healthcare, autonomous driving and security. The dependence on large labeled datasets for supervised learning poses significant challenges, particularly in specialized fields where the labeled data is scarce and expensive to obtain. Self-supervised learning (SSL) has emerged as a promising paradigm, enabling models to learn useful representations from unlabelled data by designing pretext tasks that generate pseudo-labels. SSL faces limitations in handling complex data distributions and achieving robust generalization. This paper explores hybrid self-supervised learning strategies that combine multiple SSL techniques, such as contrastive learning, masked image modeling, and clustering, to enhance image classification performance and reduce dependence on labeled data. This study proposes a comprehensive framework that integrates data augmentation, feature extraction, and hybrid learning mechanisms, evaluated on the CIFAR-100 dataset. The experimental results demonstrate that hybrid SSL approaches achieve significant improvements in performance. The combination of SimCLR and masked image modeling (MAE) achieves a Top-1 accuracy of 77.8% on the clean test set and 71.4% on the domain-shifted set, and self-distillation with contrastive learning (DINO) achieves the highest Top-1 accuracy of 78.4% on the clean test set and 72.1% on the domain-shifted set. Advanced data augmentation techniques, such as CutMix and RandAugment, additionally enhance model robustness, with SwAV (contrastive clustering) achieving 76.5% Top-1 accuracy on the clean test set and 70.1% on the domain-shifted set. The findings highlight the effectiveness of hybrid SSL methods in addressing the challenges of limited labelled data, offering valuable insights for future research and applications in image classification. 2025 Seventh Sense Research Group. -
Enhancing Music Emotion Recognition with LSTM: Evaluating Various Embedding Techniques
The study investigates the application of Long Short-Term Memory (LSTM) networks for emotion classification in music lyrics. It focuses on the comparative effectiveness of various word embedding techniques. It evaluates the performance of static embeddings (GloVe, Word2Vec, FastText) versus contextual embeddings (BERT, Distil BERT) across three datasets: MER Lyrics, Mood Lyrics, and Combined Lyrics. Additionally, the study examines the role of stylistic and content-based features in enhancing classification accuracy. The results demonstrate that contextual embeddings considerably outperform static embeddings, achieving accuracy rates of up to 98% compared to 60% for static approaches. Moreover, combining multiple lyric datasets leads to improved model generalization. The findings show the potential of transformer-based models for advancing music emotion recognition. Future research will focus on optimizing large embedding models using techniques such as pruning, quantization, and distillation to enhance computational efficiency. 2025 Seventh Sense Research Group. -
UWB Monostatic RADAR-Based Heartbeat Monitoring in an Autonomous Vehicle
Monitoring a driver's physiological state in real time is vital for enhancing road safety by detecting fatigue, medical emergencies, and enabling future health-intervention systems in autonomous vehicles. Ultra-Wideband (UWB) impulse radio monostatic Radar emerges as an attractive alternative due to its ability to perform non-invasive and highly sensitive detection of vital signs, including respiration and heart rate, through obstacles such as clothing or car seats. This paper presents a radar setup located in the seat, which propagates a UWB signal through human tissues from the back side of the driver up to the heart location. The transmitted and reflected UWB signal and antenna reflection coefficient S11 parameter are analysed to detect the heart rate for a heartbeat-induced heart model. Various UWB pulse types and their spectral characteristics are analysed to ensure efficient energy transmission within the FCC mask safety constraints. Time-domain analysis of the transmitted and received pulses reveals clear heartbeat analysis with minimal distortion, achieving accurate heart detection rates. Reflected-pulse analysis shows clear differences in amplitude between systole and diastole for normal and abnormal heart-radius conditions, allowing reliable detection of heart states. Time-of-flight and range estimation help in tracking the heart-wall movement accurately. FFT-based analysis of the time-varying S11 parameter estimates the heart rate, confirming precise non-invasive heartbeat detection through the thorax. . This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) -
Precision Food Crop Mapping Using Deep Neural Networks and Improved Dipper Throat Optimization Techniques
In recent times, the use of Remote Sensing (RS) data obtained from Unmanned Aerial Vehicles (UAVs) has gained significant popularity in crop classification tasks, including crop mapping, yield prediction, and soil classification. The classification of food crops utilizing RS Imageries (RSI) is a major application of RS tools in crop growing. Meeting the conditions for investigating these data requires more difficult approaches, and Artificial Intelligence (AI) technologies offer the mandatory support. Because of the variation and division of crop planting, archetypal classification methods have fewer classification outcomes. This manuscript focuses on the design and execution of a Leveraging Enhanced Dipper Throat Optimization Algorithm with Dipper-Inspired Precision Classification for Remote-sensed Optimized (DIP-CROP) Processing methodology. The drive of the DIP-CROP algorithm is to classify distinct types of crops that exist in remote sensing. At first, the DIP-CROP model applies image processing using the Sobel Filter (SF) to eliminate the noise. Next, the presented DIP-CROP technique takes place SqueezeNet model is employed for the feature extractor. To classify the food crop types, the DIP-CROP approach utilizes a Multi-Head Attention-based Bi-directional Long Short Term Memory (MHA-BiLSTM) algorithm. For hyperparameter tuning of the MHA-BiLSTM classifier, the Enhanced Dipper Throat Optimization Algorithm (EDTOA) will be applied in this work. The optimization process utilizes Levy flight distribution, which is known for its faster convergence due to efficient exploration of the search space. Levy flights can be used to take larger steps in exploration, which prevents getting stuck in local minima and accelerates convergence. The performance of the DIP CROP method is examined experimentally using a benchmark database. Experimental results affirmed the superior solution of the DIP-CROP algorithm over existing methods. 2026 Seventh Sense Research Group. -
Censored Regressive Canonical Optimized Convolutional Deep Belief Classifier For Hate Speech Detection in Online Social Network
Social networking uses internet-based platforms to facilitate users to make connections with others and share various forms of content, including text, images, videos, and links. Social networking services are mainly used for non-social interpersonal communication. Many approaches have been developed for hate speech detection, but they still face significant challenges, particularly in classifying text into multiple labels accurately and in a timely manner. For accurate hate speech detection in social networks, a Censored Regressive Canonical Optimized Convolutional Deep Belief Classifier (CRCOCDBC) model is developed. The objective of the developed CRCOCDBC is to detect multi-class hate speech with minimal time and error rate. Comparative analysis shows improved performance in terms of minimum error and higher authentication accuracy and precision than other well-known methods. 2026 Seventh Sense Research Group. -
A Deep Learning-Based BCI System for Emotion Classification Using EEG Signals
Electroencephalography-based Brain-Computer Interfacing (EEG-BCI) technologies allow for effortless interaction between external hardware and the human brain through monitoring its electric signals. These systems rely on EEG recordings, which provide non-invasive and real-time neural information through electrodes placed on the scalp. To advance emotion-recognizing efficiency and accuracy, this study proposes a deep learning-based method that can extract valuable temporal and spatial information from EEG signals. The proposed model includes the use of a Graph Convolution Network (GCN) for learning spatial relationships between different EEG channels to model the data in graph form and gain features through that modelling. A Convolutional Autoencoder (CAE) is then used to compress data to low dimensions and to reconstruct it so that major features are not ignored. Furthermore, the model uses an Attention-based Bidirectional Gated Recurrent Unit (ABiGRU) for temporal classification, which can emphasize the most important time steps in both backwards and forward directions. Two standard datasets are employed to test the developed approach. The DEAP dataset is used for emotion recognition with a binary response, and SEED is used with multi-class classification. The model attains great results of 98.12% accuracy on DEAP and 97.58% on SEED datasets. The very high performances show the efficacy of the model for decoding emotional states from EEG signals and very strong potential for real-time emotion recognition in affective computing and BCI. 2026 Seventh Sense Research Group. -
A Comparative Analysis of Various Soft Computing Techniques for Indian Stock Market Prediction
Soft computing techniques have been increasingly used for stock market analysis in the past few years because they can capture nonlinear aspects which traditional econometric models do not adequately capture. With different techniques like Artificial Neural Networks, Deep Neural Networks and Stacked Autoencoders available, in this paper, the author tries to determine which of the above methods can model the Indian stock market with higher accuracy. In this study, high-frequency data from Nifty 50 is used, and various feature selection techniques such as PCA and linear regression are used for each of the above machine learning models to predict the Nifty 50 data. Finally, all predictions from the different techniques are compared with the actual index movement and the best method for Nifty 50 is suggested. 2025 Seventh Sense Research Group -
A Comparative Analysis of L1, L2, and L1L2 Regularization Techniques in Neural Networks for Image Classification
The research examines how L1, L2, and L1L2 weight regularization methods affect neural network performance, generalization, and sparsity using the CIFAR 10 dataset. A Convolutional Neural Network (CNN) trained with the same environment for each regularization method to evaluate test accuracy, weight sparsity, and computational speed. The study shows that L1 regularization produces sparse weights, which makes models more interpretable, and L2 regularization helps prevent overfitting while improving model generalization. The combination of L1L2 regularization enables individual image classification methods to reach test accuracy. The results indicate that the weight regularization plays a vital role in creating neural networks that are both stable and efficient. They are interpretable, and L2 regularization improves generalization and reduces overfitting. The combined L1L2 regularization achieves the balance between sparsity and performance, leading to higher test accuracy compared to individual techniques for image classification. The research results demonstrate that weight regularization stands as an essential factor for Creating Neural Networks that are robust, efficient, and interpretable, thus helping to enhance Deep Learning model performance. 2025 Seventh Sense Research Group. -
Predictive modeling of mechanical behavior in waste ceramic concrete using machine learning techniques
This study identifies the critical demand for a certain approach that aims to predict and ascertain the mechanical behavior of concrete admixed with waste ceramic, a method to overcome and mitigate the related environmental challenges as it pertains to the construction field. Concrete modification with ceramic wastes has received significant attention due to its potential improvement in sustainability. The developed predictive models on waste ceramic concrete (WCC) involved the use of advanced machine learning techniques such as Artificial Neural Network (ANN) and Light Gradient Boosting Machine (LightGBM). Experimental datasets were formulated based on 5% and 20% variability of ceramic waste percentages as input variables for training and testing data for validation of the proposed model. In each case, iterative training improved model performance, with the ANN showing moderate predictability (R = 0.70 and 0.67) and LightGBM demonstrating stronger accuracy. Predictive values ranged between 1.02 MPa and 0.12 MPa for compressive and splitting tensile strengths and had R values of 0.70 and 0.67 for the ANN model, respectively. The established findings will lead to a dependable framework for assessing and improving the performance of ceramic waste-modified concrete. In this regard, these findings have reinforced the potential of machine learning in developing sustainable construction practices. This paper is of value to engineers and decision-makers within the construction industry, providing an informed choice towards environmental sustainability and better risk management. Kamal Upreti et al. -
Does financial inclusion and ICT influence the economic growth: an evidence from select emerging economies using panel data estimates
Information Communication Technology (ICT) 's contribution to financial services and economic growth is extensively researched. With technological development, the financial systems are converging with ICT platforms, thereby leading to digital financial systems, creating opportunities that bridge the gap between affluent and disadvantaged sections in emerging economies. In this study, we employ the panel data estimation method to evaluate the impact of ICT in providing financial services that would lead to sustainable economic growth in 12 emerging economies from 2010 to 2024. The findings emphasize the critical role of digital financial services and ICT infrastructure in boosting financial inclusion and driving inclusive economic growth. The Random Effect model has inferred that there exists a moderate effect of fixed telephone lines and mobile cellular subscribers on GDP by simplifying digital applications, integrating regional languages. By ensuring secure systems in places, the regulators and service providers can contribute to the sustainable economic growth of both the country and the underserved communities. Shylaja H N et al. -
A Sustainable Business Model for Converting Construction and Demolition Waste to Wealth
India's rapid urbanisation necessitates a planning approach that ensures the sustainability of its cities through efficient waste management. This swift urban growth has significantly accelerated modern construction and demolition of older infrastructure or structures within Indian cities. C&D (Construction and Demolition) waste is accountable for approximately 30 percent of urban municipal waste within metropolitan areas. Managing C&D waste and transforming it into valuable resources presents considerable challenges for all urban local bodies (ULBs). Recycling C&D waste offers dual benefits: it reduces pressure on the extraction of virgin construction materials and helps mitigate environmental pollution. Recycled C&D waste can produce various valuable products, including aggregates of different sizes, manufactured sand, paver blocks, concrete bricks, double-tee precast panels for boundary walls, manhole covers, water tanks, and more. These products are durable and eco-friendly building materials that contribute to the conservation of natural resources. However, a sustainable business model is essential for understanding the volume of C&D waste produced and for addressing current challenges and opportunities at the city, regional, and state levels. The current research aims to gather information about the overall scenario of C&D waste management procedures in India, relying on secondary resources. It proposes a sustainable business model for C&D waste handling that transforms this specific waste into a valuable resource, identifying possible advantages and the resource efficiency of recycled items. permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. -
Workforce Transformation and Value Creation in the Era of Industry 4.0, 5.0 and 6.0: Challenges and Enablers
Industry 4.0, Industry 5.0, and Industry 6.0 have been empowered by the acceptance of several advanced technologies, including the Internet of Things (IoT), artificial intelligence, robotics, and human-centric innovation in releasing industries. Comprehensive studies that can include barriers as well as enablers are difficult to conduct. This study employs a mixed-methods research approach, integrating of AHP (Analytic Hierarchy Process) and TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) to evaluate key challenges and enablers in workforce transformation. The Findings indicate that leadership vision, digital investment, and employee up-skilling play a crucial role in transformation navigation. Additionally, automation and AI adoption present both opportunities and challenges for workforce adaptability. The study provides strategic insights for organizations to enhance their workforce resilience, competi-tiveness in the evolving industrial landscape. The study offers actionable advice for businesses, policymakers, and educators, and suc-cessfully adaptation the paradigm and sustainable development in the modern era. Authors. -
Structural Modelling of Homebuyer Behavior in The Residential Housing Market
Rapid urbanization has significantly increased the demand for residential housing, underscoring the need for real estate retailers to comprehend the factors that drive homebuyers' purchasing behavior. This study employs a Structural Equation Modelling (SEM) framework to examine the critical determinants of purchase intention within the residential real estate sector. Key constructs analyzed include housing attributes, financial and economic considerations, location and service quality, environmental concerns, brand image, and information sources. The SEM analysis elucidates the strength and significance of the relationships among these variables, providing a comprehensive understanding of consumer decision-making processes. The findings yield actionable insights for real estate practitioners, offering guidance for the development of targeted marketing strategies, enhanced customer engagement practices, and improved service delivery. By aligning their offerings with buyer expectations, real estate firms can foster stronger client relationships and enhance their competitive positioning in a dynamic housing market. Authors. -
Predicting financial asset prices with neural network: a comparative study of neural networks effectiveness in financial decision-making
Investing requires deep knowledge of complex financial markets, making it incredibly tedious to predict inflation and deflation. Predictive conventional models like ARIMA and GARCH do not accurately capture the non-linearity and volatility presented in financial datasets. This research examines the different forms of predictive assets, real estate, stocks, commodities, bonds, and cryptocurrency using Long Short-Term Memory (LSTM) Neural networks. The primary focus of this research is to assess the valuable prediction capabilities of LSTM across assets and its integration with financial decision-making. According to the empirical results, deep learning LSTM models give better outcomes with equities and gold, with the R2 indicator reaching over 99% alongside a low RMSE. LSTMs had an over 100% MPE prediction error rate for other assets during the test phase, making it harder to predict intensely volatile assets. The model's verification transfers residual autocorrelation, showing that it can enhance forecasting performance with detailed macroeconomic indicators and sentiment analysis data. Studies show that LSTMs are effective in high-frequency markets with non-linear price changes, but they require special attention to balance interpretability and overfitting. Despite the progress that has been achieved in utilizing neural networks for financial forecasting, hybrid models integrated with XAI are recommended to improve efficiency and real-world applicability. These results contribute to the growing domain of AI-powered finance by offering additional means for many investors, analysts, and decision-makers who wish to utilize data for market speculation. Dr. Aishwarya Nagarathinam et al. -
Sustainable Marketing and Green Finance: Integrating ESG Metrics into Financial Reporting and Strategic Branding
As sustainability gains significance in the global business landscape, an increasing number of companies are adopting Environmental, Social, and Governance (ESG) frameworks to enhance transparency and strengthen stakeholder relationships. This study looks at how using ESG metrics in branding and financial reporting affects the creation of long-term corporate value. The main goal is to find out how combining ESG initiatives with marketing plans and financial disclosures affects brand equity and financial credibility. There are both qualitative and quantitative parts to the study. This involves a qualitative analysis of ESG reports from 150 multinational corporations and the utilization of quantitative regression methods to examine the impact of ESG integration on brand performance and financial metrics. The Global Reporting Initiative (GRI) set the rules for ESG scores. Return on Assets (ROA) and Tobin's Q were the most important financial measures. The results show that companies with high levels of ESG integration saw a 12.4% increase in ROA and a 0.38 average increase in Tobin's Q compared to companies with low levels of ESG activity. Both changes were statistically significant at p < 0.01. Survey data also showed that companies that closely linked their ESG disclosures to their branding had a brand trust index that was 22% higher. These results show that strategically branded, ESG-focused reporting not only improves financial performance but also makes consumers feel better about the company. Ultimately, the study offers a framework for integrating ESG metrics into financial and marketing strategies, emphasizing ESG's function as both a moral obligation and a source of competitive advantage in the context of responsible capitalism. Sireesha Nanduri et al.