
The rising use of System-on-Chip (SoC) technology in the fields of embedded vision, surveillance, biomedical imagingand portable electronic devices has produced a growing demand for high performance and low power image preprocessing systems. Random-valued impulse noise is one of the many variables that degrades images and severely degrades the quality of the collected binary image data, which badly affects future image analysis and processing operations. This work proposes an efficient SoC based architecture to remove the random valued impulse noise from the collected binary data in order to overcome this problem. The decision-tree-based detector to detect the noisy values and employs an effective design to locate the edge is proposed in this design. Decision tree based hybrid integrated de-noising method (DTHIDM) is used as proposed methodology for this project to implement efficient de-noising process. If there are noisy values, edges, or blocks in this region, the distribution of the values is different. Therefore, we determine whether current pixel is an Seclusion point by observing the smoothness of its surrounding pixels. The noise may seriously affect the performance of image processing techniques. Hence, an efficient denoising technique becomes a very important issue in data processing.
Student talent enhancement is highly concentrated in today’s educational system. Students are getting distracted by several activities, and their concentration is out of education. Therefore an effective model is needed to enhance to students’ talent by performance several activities on daily basis to them. To address this issue, a new model created the integrated DL model with Hybrid Optimal Attention in Edge-Cloud network (IDHOEC) framework units several cutting-edge technologies to detect the student talent. To achieve these, Edge-Cloud-IoT systems equipped with Raspberry Pi and Arduino sensors are used for managing data in real-time. The cloud setup is in charge of image storage, handling databases, and displaying information through LXD containerized services. Student monitoring is done through YOLO-based object detection. A bio-inspired approach not only increases the effectiveness of the model but also makes its decisions more understandable. The use of a Bayesian-optimized LSTM (BO-LSTM) model leads to better student performance prediction. Capsule Network (CapsNet) and dynamic routing help preserve spatial relationships when features are being represented. Deep Neural Network (DNN) are capable of learning nonlinear mappings and after that, they are optimized using gradient descent. A Modified Aquila Optimization (MAO)-optimized attention-based CapsNet enhances feature extraction and accuracy.
Human Resource Auditing and Disclosure Systems (HRADS) are increasingly recognized as vital mechanisms forenhancing accountability, compliance, and strategic alignment in organizations. This study investigates the determinants and consequences of HRADS adoption in the Indian IT sector by integrating Institutional Theory and Contingency Theory. Drawing on survey data from 300 respondents across multiple departments, the study employs structural equation modeling to test thirteen hypotheses, supplemented with ANOVA and mediation analyses. The results demonstrate that top management support, HRIS maturity, auditor competence, resource availability, and external forces such as regulatory mandates and client pressures significantly influence HRADS adoption, scope, and impact. Further, certifications, data privacy readiness, and a culture of organizational learning strengthen audit depth and effectiveness. Interaction effects, including remote work intensity and client certification pressures, also exhibit significant influence. ANOVA results reveal notable departmental differences in HRADS adoption (F = 5.92, p < 0.001), underscoring the heterogeneous implementation of audit practices. Mediation analysis confirms that audit scope serves as a critical conduit linking data privacy readiness to organizational impact, validated through bootstrapped indirect effects. Collectively, these findings highlight that HR audits extend beyond internal governance mechanisms, functioning as institutional responses to legitimacy pressures while also serving as contingent tools for aligning strategy with context-specific organizational needs. The study contributes to both theory and practice by providing an integrative perspective on HR auditing, offering implications for regulators, practitioners, and policymakers aiming to strengthen human resource governance in dynamic, compliance-driven environments.
Credit Card Fauld Detection in Banking application technological advancements complicate cybercrime and hinder credit card fraud detection. Traditional methods are ineffective. A company developed an event-based temporal graph (ETG) to detail transactions among users, cards, devices, and accounts. This review examines machine learning algorithms for detecting financial fraud amid rising digital transactions. We examine many forms of fraud, such as money laundering, insurance fraud, credit card fraud, and financial statement fraud, as well as the unique detection difficulties associated with each new model Credit Card Fauld Detection Through Deep Learning Approaches with Data Analytics (CCFDDA). A method for credit card fraud detection that integrates deep learning and traditional machine learning with data preprocessing. Using a Kaggle dataset of 284, 807 transactions (492 fraudulent), we perform Principal Component Analysis (PCA) for feature anonymization and address class imbalance. Our analysis identifies key fraud indicators through statistical methods, and we enhance data completeness by removing incomplete rows, improving model learnability. In the Feature Normalization phase, Min-Max scaling normalizes numerical attributes to the range of 0 to 1 to mitigate scale bias. The cleaned transaction data is represented using overlapping sliding windows to maintain temporal dependencies. The machine-learning fraud detection component evaluates several models, including Decision Trees (DT), Random Forests (RF), Support Vector Machines (SVM), XGBoost, CatBoost, and Logistic Regression (LR), to identify fraudulent transactions and addresses class imbalance through various classification and ensemble techniques. Softmax assigns class probabilities, while a class-weighted loss function addresses class imbalance. The architecture employs machine learning, statistical analysis, structured preprocessing, and temporal graph-based deep learning for fraud detection. Its Deep Learning module utilizes a Dynamic Graph Neural Network (DGNN) for time-dependent classification and message passing, enabling it to model changing transaction interactions over time for effective fraud detection.
As cyberattacks are increasing in today’s digital world, especially Distributed Denial of Service attacks, securing network resources has become crucial for both individuals and network administrators. Here comes the need for a reliable and trusted Intrusion Detection System, which plays a significant role in determining and detecting threats by monitoring network traffic and alerting the network administrator. Therefore, we have designed and evaluated a deep learning-based IDS that focuses on DDoS network attacks using a popular dataset. This CICDDoS2019 dataset contains a wide range of network traffic patterns. To strengthen the model’s performance, we have used the Shapley method for feature selection along with the Hotelling T2 test to choose the top contributing set of features, which contribute the most in deciding whether there is an attack or not. The feature selection method reduced the original 87 features to the 21 most predictive attributes. After selecting the features, we used a dense multilayer perceptron (MLP) — a deep learning model well-known for recognising complex data patterns. The model was trained and tested on the optimal dataset to differentiate benign and DDoS network attacks. The model performance is evaluated using metrics like accuracy, recall, precision, balanced accuracy, and F1-score. The results show that combining Shapley with the Dense MLP model achieved 99.75±0.020% accuracy with consistently high precision (99.72±0.037%), recall (99.78±0.012%), and ROC-AUC (>0.995) across datasets on three independent runs. The proposed framework also demonstrated highly accurate detection of DDoS attacks, maintaining a low false-positive rate while effectively distinguishing malicious traffic from legitimate network activities. This study shows the impact of meaningful feature selection in enhancing the predictive capability of deep learning models for intrusion detection. The results demonstrate that integrating meaningful feature selection with deep learning strengthens IDS performance offers a practical, interpretable, and computationally efficient solution for real-time DDoS detection.