Browse Items (16481 total)
Sort by:
-
An Algorithm for Cybersecurity Threats Detection in the Internet of Things using Deep Learning Approach
We perform research to develop a combined deep learning algorithm that enhances security threat detection within the Internet of Things networks. The resource variations across IoT devices create obstacles for Traditional Intrusion Detection Systems (IDSs) regarding their scalability and adaptability elements. This study explores the application of Bidirectional Recurrent Neural Networks and Long Short-Term Memory networks, which are trained on Traffic data records from NSL-KDD, a widely recognized benchmark dataset. It's a secondary dataset. This dataset is preprocessed and features are engineered to be optimized for sequential pattern recognition and handling of long-term dependency. Experimental results validate the achievement of a cross-validation accuracy of 93.40%, F1 is 91.62% and precision is 90.42%, which is greater than the individual models, such as CNN, BiRNN, or LSTM. The stacking Models Bi-RNN sequential learning and LSTM dependency retention makes the system perform better at threat classification along with elevated detection accuracy for IoT-related security issues like DoS, Probe, R2L, and U2R. The consistent performance of the model through this validation split provides evidence that the system can effectively handle IoT cybersecurity threats. 2025 IEEE. -
An Explainable AI-Driven Deep Learning Algorithm for Heart Disease Detection in Healthcare
The application of preprocessed Kaggle data serves as a subject of analysis to investigate heart attack prediction capabilities through machine learning models. The research examines performance outcomes of five algorithms which consist of K-Nearest Neighbors (KNN), Random Forest, Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost) and Convolutional Neural Networks (CNN). Random Forest together with XGBoost proved as the most accurate machine learning models when used for cardiovascular risk assessment. The researchers built a hybrid structure of CNN and SVM because it improved both data classification and feature extraction processes for better prediction outcomes. The training and evaluation process of models encountered difficulties because of overfitting along with high computational expenses and problems regarding optimal hyperparameter settings. The research stresses that explainable AI (XAI) methods should integrate into systems to enhance model interpretability and achieve trust from clinical professionals. Future initiatives seek real-time patient monitoring and innovative interpretability systems for heart attack prediction to enable person-specific diagnoses and optimal clinical choices in medical fields. 2025 IEEE. -
A Novel Approach to Packet Dropping and Malicious Attack Detection using Ensemble Techniques
Packet-dropping attacks interrupt data transfer while damaging security protocols, which create a threat to wireless Sensor Networks and Mobile Ad Hoc networks. This paper examines packet-dropping detection methods as well as security attack identification since these threats represent significant risks to networks such as Wireless sensor networks and Mobile Ad Hoc Networks. The research paper utilized a dataset from Kaggle for network traffic analysis, which classified packets through their behaviors as either abnormal or normal. The detection employed a stacking classifier with logistic regression as the meta-classifier and Support Vector Machine, Gradient Boosting, and K-Nearest Neighbour as its main constituents. The analysis model showed high detection rates for packet-dropping incidents, reaching 93.5%, and for malicious attacks, reaching 98.2%, based on the experimental test results. The obtained data shows that stacking models show stable reliability levels above traditional approaches. Ensemble learning proves effective for discovering cyber threats through results that reduce the number of incorrect detections. The stacking classifier functions as a dependable framework for developing security measures required to protect computer networks from modern-day threats. 2025 IEEE. -
Enhancing Malware Detection Through Hybrid Deep Learning Techniques
The detection of malware needs superior methods than basic signature detection because it remains vital to cybersecurity. This research examines malware classification through the deep learning approach by analyzing Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU) and develops a new BiGRU + CNN hybrid model. The main purpose is to achieve better detection performance through reduced numbers of false alarms. The research employs executable file feature data while implementing preprocessing methods together with fivefold cross-validation validation to establish strong model reliability. Experimental findings show CNN along with LSTM and GRU attains excellent recall values yet produces elevated erroneous positive predictions. The proposed BiGRU + CNN model delivers superiority over single-model architecture as it reaches 96.06% accuracy alongside 96.13% precision and 99.92% recall and 97.99% F1-score. The obtained results show that this integration has better malware detection capabilities thereby demonstrating its potential for cybersecurity applications. 2025 IEEE. -
Ensemble Hybrid LSTM Architectures for Robust Multi-Currency Forex Forecasting
The analysis of financial time series presents a longlasting obstacle regarding currency exchange rate forecasting because volatility and nonlinearity and non-stationarity characterize currency markets. The research presents an ensemble forecasting system which combines various deep learning and hybrid predictive models such as LSTM and GRU-LSTM and CNN-LSTM and Attention-LSTM and XGBoost-LSTM for scalable integration. The ensemble methodology follows a dynamic weighted averaging technique which bases its priority on assigning weights through the reciprocal calculation of Mean Squared Errors from individual models to identify accurate forecasters. A representative study based on the EUR/USD exchange rate took place as part of extensive evaluations that spanned various currency pairs. The standalone XGBoost-LSTM model proved most effective in terms of MSE and R2 values at 0.000088 and 0.9778 respectively. The ensemble model proved to be highly robust and generalizable through its outcomes which produced an MSE of 0.000142 along with MAE of 0.009204 and R2 of 0.9643. The ensemble approach stands as an effective and reliable method to increase both stability and predictive power of forex forecasting systems. The conceptual structure offers sound potential applications for algorithmic trading as well as financial risk management and multi-currency strategic decision-making systems. 2025 IEEE. -
Artificial Intelligence in Banking Security-Technical Innovations and Challenges
The accelerating adoption of artificial intelligence (AI) technologies in the banking sector has introduced transformative possibilities for enhancing security frameworks against increasingly sophisticated cyber threats. This research investigates the technical innovations driven by AI, such as machine learning algorithms, biometric authentication systems, and natural language processing, and their impact on improving fraud detection, cybersecurity monitoring, and compliance automation. The paper identifies how AI systems, through real-time analysis of large-scale transaction data, can locate abnormal behavioral patterns and respond proactively to potential threats, significantly reducing human error and response time. A detailed analysis of the current literature reveals a significant research gap in integrating explainable AI, secure data governance frameworks, and scalable models suited for diverse banking environments. The outcome of this research highlights the need for a balanced approach that fosters technological innovation while addressing regulatory compliance, ethical concerns, and operational constraints, paving the way for a secure and intelligent banking infrastructure. 2025 IEEE. -
Deep Learning-based Cybersecurity Framework for IoT Environments
The article "Cyber Security In IOT using deep learning Approach"presumably states the pressing need for increased cybersecurity within the rapidly growing Internet of Things (IoT) paradigm. It stresses the distinct problems brought by IoT settings, such as distributed systems and heterogeneous devices, that make the old methods of security inoperable. The article we underscores the promise of deep learning and AI as novel solutions for identifying and foiling cyberattacks, but also notes the emergence of adversarial AI employed by cybercriminals. In addition, it urges proactive cybersecurity measures and ongoing surveillance to counter changing threats, especially with IoT networks becoming increasingly complex with applications in smart cities and other industries. The abstract can conclude by emphasizing the need for continued research into AI-based cybersecurity solutions to guarantee effective protection against emerging threats. 2025 IEEE. -
Deep Learning-based Cybersecurity Framework for IoT Environments
The article "Cyber Security In IOT using deep learning Approach"presumably states the pressing need for increased cybersecurity within the rapidly growing Internet of Things (IoT) paradigm. It stresses the distinct problems brought by IoT settings, such as distributed systems and heterogeneous devices, that make the old methods of security inoperable. The article we underscores the promise of deep learning and AI as novel solutions for identifying and foiling cyberattacks, but also notes the emergence of adversarial AI employed by cybercriminals. In addition, it urges proactive cybersecurity measures and ongoing surveillance to counter changing threats, especially with IoT networks becoming increasingly complex with applications in smart cities and other industries. The abstract can conclude by emphasizing the need for continued research into AI-based cybersecurity solutions to guarantee effective protection against emerging threats. 2025 IEEE. -
Geochemical Data Exploration using Machine Learning Methods
This study introduces a novel ensemble model combining Support Vector Machine (SVM) and Gradient Boosting algorithm (GBC). The model's performance is compared with the two single layered model namely K-Nearest Neighbors (KNN) and Gaussian Naive Bayes (GNB) on a publicly available dataset. Further, Performance is measured using standard metrics such as accuracy, precision, and recall. To have the excellence in detection of types of rocks based on its properties this research explores the stacking approach, contributing in the field of geological studies and also for future exploration making it effective and efficient in identification of mineral deposits. 2025 IEEE. -
Credit Card Fraud Detection with ADASYN Oversampling and SHAP-based Interpretability: A Comparative Ensemble Approach
Credit card fraud continues to be a significant threat to financial systems, exacerbated by the highly imbalanced nature of transaction datasets and the opaque decision-making of complex machine learning models. This paper proposes a hybrid fraud detection framework that integrates Adaptive Synthetic (ADASYN) oversampling to address class imbalance and SHAP (SHapley Additive exPlanations) to enhance model interpretability. Five machine learning classifiers Logistic Regression, Random Forest, XGBoost, LightGBM, and Multilayer Perceptron - are evaluated on the widely used Kaggle credit card fraud dataset. ADASYN significantly improves the minority class representation in the training set, enabling models to achieve higher fraud recall without overwhelming false positives. Among the models tested, Random Forest delivered the best trade-off between precision (85.7%) and recall (79.6%), achieving an F1-score of 82.5% and ROC-AUC of 0.9633. SHAP analysis provided granular insight into feature contributions, transforming black-box predictions into transparent and auditable decisions. Comparative analysis with eight state-of-the-art studies demonstrates that while recent approaches often report near-perfect results, the proposed model strikes a balance between predictive performance, computational efficiency, and interpretability qualities essential for practical deployment in financial fraud detection systems based on benchmark transactional data. The study highlights that integrating ADASYN with ensemble learning and SHAP can create a robust, explainable, and scalable fraud detection system suitable for deployment in dynamic financial environments. 2025 IEEE. -
Optimizing Disease Diagnosis and Treatment Through AI and Deep Learning Algorithms
A Primer for Cancer Center Leaders Session 2 Natural Language Processing for Biomedical Text Medical data is not only numeric but also composed of unstructured text. These algorithms listen to various medical imaging, genomic data, and electronic health records to find correlations that can predict different diseases. Using convolutional neural networks to analyze images and recurrent neural networks to process sequential data, AI systems improve diagnostic accuracy and minimize the risk of human error. Additionally, deep learning algorithms targets patient-oriented drug administration by predicting therapeutic responses of individual patients, enhancing treatment response. Incorporating AI into clinical workflows allows us to synthesize vast datasets in real-time, provide clinicians with action items, and advocate for evidence-based medicine. However, problems including data privacy, model interpretability, and the need for large, annotated datasets continue. Such solutions in the form of explainable AI and deep learning would play an integral role in promoting the usage of these technologies over a longer duration in the medical ecosystem. This work shows how AI and deep learning can open avenues that may fundamentally change disease detection and treatments, leading to improved diagnosis and treatments tailored to the individual patient. 2025 IEEE. -
A Novel Framework for Integrating Machine Learning in CSR to Accelerate Sustainability in the Indian Automobile Sector
This paper aims at determining the suitability of using machine learning in CSR in enhancing sustainability of the Indian automobile industry. Prominent automobile companies are known to be major sources of environmental pollution together with wastage of various natural resources. There is a challenge of incorporating sustainability policies in the sector due to the rising regulation and consumers' awareness. Machine learning contains new approaches to managing resources more effectively, minimizing emissions and providing transparency of goods to clients. This study scrutinizes the previous literature, outlines the machine learning-based framework system of CSR activities, and validates the applicability of the system using case studies and qualitative data. The results show that learning from data can improve sustainability at an extensive scale and that the changes are sustainable and financially advantageous. 2025 IEEE. -
Early Warning System for Engine Failure Detection in Aircraft Engines Using Machine Learning
Aviation has a problem with engine defects which are a major concern. Unforeseen causes might render them expensive on the ground and hazardous in the air. We present a system that signals when an aircraft engine is about to fail. Our AdvancedModelTrainer checks a collection of models - Random Forest, XGBoost, Gradient Boosting, LightGBM, Ridge, Lasso, ElasticNet, and a simple neural network - through a dataset of 10,000 engine cycles along with 25 engineered features. Hyperparameter tuning and Remaining Useful Life (RUL) metrics help to select the top two (Gradient Boosting and XGBoost, RMSE 39.99, R2=0.7715). A complete MLOps structure keeps an eye on the drift, initiates the retraining process, and sets up dashboards that are user-friendly for the mechanics. The system has detected on 1,433 new engines, 1,126 were classified as Safe, 106 as Warning, and 201 as Critical, which is indicating the coverage of 93.44The dataset used was completely anonymized in order to safeguard sensitive operational data and to not conflict with the aviation data privacy regulations. 2025 IEEE. -
Deep Learning Ensemble for Neuro-Oncological Diagnostics: Fusion of VGG-16 and Convolutional Neural Network Architectures
Brain tumors are potentially fatal, prompt and accurate diagnosis is essential to appropriate treatment and management. MRI is a key method for locating tumors in the brain. This study introduces a HYBRID deep learning for binary classification of brain tumors, combining a pre trained VGG16 model with tailored CNN and Neural Networks. The fusion of these models is done via feature concatenation followed by a common classifier. This fusion helps in capturing both high-level abstract and task-specific features critical for classification. To help minimize overfitting and improve generalization, the models are subjected to rigorous data augmentation including rotation, zooming, and horizontal flipping, normalization, and resizing of images to 150 150 pixels. All models are trained and validated using the same data splits. Performance is determined by accuracy, training and validation loss, confusion matrices, and visualization with Matplotlib plots and Plotly which provide a vivid insight into the models. Experiments are conducted to determine the different model performances and the hybrid model attained an accuracy of 98.14%, which was higher than the standalone VGG16 (93%), CNN (91%), and NN (88%) models. 2025 IEEE. -
Design and Implementation of a High-Speed Level Shifter at 45nm, 90nm, and 180nm Technology Nodes using Cadence
In this work, a CMOS inverter-based level shifter in Differential Cascode Voltage Switch Logic (DCVSL) is constructed and its operation is investigated. The width and length variations of transistors at three technological nodes 45, 90, and 180 nm are compared based on the circuit behaviour. A critical analysis of the impact of supply voltage scaling on NMOS and PMOS transistors is also presented. There is also a comparison of the effects of transistor widths and lengths, as well as supply voltage variations of 1.8V, 1.5V, and 1.0V, on circuit performance. Additionally, this study compares wavelength variation and its impact on device attributes. Dynamic power, static power, energy, and delay are evaluated at the transistor and direct current levels. The Cadence Virtuoso simulation tool illustrates the variations in inverter performance under various scaling conditions. The results demonstrate that careful optimization of transistor dimensions and supply voltage can significantly enhance the performance and power efficiency of the level shifter, providing valuable insights for low-power, high-speed VLSI applications. 2025 IEEE. -
Wheat Disease Diagnosis using Transfer Learning on Convolutional Neural Networks
Wheat disease identification is essential for agricultural output and food security. Traditional diagnostic approaches are slow, ineffective, and need expert assistance, restricting their ability to grow in agriculture. Suggested innovative diagnostic method uses transfer learning on convolutional neural networks (CNNs) to effectively identify and classify wheat leaf diseases. To increase model predictions, high-quality image datasets from open-access platforms are normalised, resized, and augmented. The proposed CNN model performed best with 98.90% accuracy, 98.87% precision, and 98.80% recall. Transfer learning improved model performance by recycling knowledge from pre-trained CNN architectures, reducing training time and enhancing feature extraction. The results show improved precision as well as strength over standard methods and before. This technology helps farmers and agricultural professionals make timely disease management and crop management decisions. To improve disease recognition, future study may use a wider dataset range and other CNN designs. 2025 IEEE. -
Charting the Complexity of Diabetes Risk using Network-based Exploration of Nonlinear Interactions
Diabetes mellitus is a global health challenge shaped by complex clinical, demographic, and socioenvironmental factors. Traditional linear models often overlook the non-linear dependencies that drive diabetes risk. This study adopts a systems-thinking approach by integrating mutual information (MI)-based network modeling with machine learning to improve prediction, interpretability, and fairness. Using a nationally representative CDC dataset, we build a weighted undirected network where variables are nodes connected by MI-derived edges. Centrality analysis identifies age, HbA1c, and BMI as key hubs. Community analysis reveals clinical, demographic, and racial modules, reflecting the multidimensional nature of diabetes risk. These network insights inform feature selection for training logistic regression, random forest, and XGBoost models. XGBoost achieves the highest accuracy (95.3%) and AUC (0.939), while logistic regression offers the best calibration (Brier score = 0.045), enhancing clinical usability. Subgroup analysis shows stable predictions across racial groups, supporting fairness. This integrated framework uncovers latent, non-linear associations and offers a robust, interpretable, and equitable tool for precision diabetes risk modeling. 2025 IEEE. -
Predicting Financial Market Volatility Using Regression and Machine Learning Techniques
In standard Simple Linear Regression (SLR), one of the major assumptions is that the error terms have constant variance (homoscedasticity). However, this assumption is frequently violated in many real-world datasets, resulting in inefficient estimates and reduced predictive accuracy. To overcome this shortcoming, we propose a hybrid modeling platform that combines SLR with statistical and machine learning methods. The approach starts with SLR to identify the main linear relationship. Whenever residual diagnostics report the presence of heteroskedasticity, an Autoregressive Conditional Heteroskedasticity (ARCH) model is used to estimate time-varying variance. Such estimated variances are utilized in a Weighted Generalized Least Squares (WGLS) model, which stabilizes the error structure. Finally, to capture any remaining nonlinear patterns, an Artificial Neural Network (ANN) is applied on the residuals of the WGLS model. By layering these techniques, the hybrid framework improves both stability and predictive power. Simulation studies and empirical tests on Apple Inc. stock data confirmed that the hybrid framework yields reduced MAE and RMSE values and greater explanatory strength than individual approaches. 2025 IEEE. -
Face-Based Kinship Verification using Deep Embeddings for Low-Cost Health Record Linkage
Precise linkage of health records is essential for continuity of care, reducing duplicate health records, and accurately documenting family medical histories. Genomic testing offers the evidence-based biological 'gold standard' for verifying kinship; however, access to testing is either impossible or unavailable in most low-resourced environments due to prohibitive costs, long timelines, and/or lack of infrastructure. This study provides a low cost and interpretable pipeline for kinship verification in the form of Siamese deep embeddings. The processed facial image embeddings produced by a ResNet-18 backbone using 256-dimensional and L2-normalized embeddings, are then compared using cosine similarity. A validation-based calibration process selects the logit polarity and decision threshold that support stable deployment decisions. Grad-CAM visualizations can be interpreted frame-by-frame and allow for pair-specific attributions of faces that were more relevant or important in decisions of similarity. In experiments on the Families in the Wild (FIW) dataset (family-disjoint splits), we report ROC-AUC of 0.834, target balanced accuracy of ?0.88, with similar precision, recall, and specificity. The confusion matrices also illustrate a near symmetric distribution of errors by family and both Grad-CAM explanations highlight how the model came to a decision for true cases and hard cases. The above results illustrate how we can deploy a lightweight, explainable, and face-based kinship verification pipeline on a CPU-only system. Our study therefore provides a feasible assistive tool for health record linkage where genomic validation is not possible. 2025 IEEE. -
Infrared Eye Tracking: Unlocking Communication Pathways for Coma Patients
Eye tracking technologies have emerged as a groundbreaking technology in assessing and facilitating communication in patients with disorders of consciousness, including patients in a coma. Traditional methods of diagnosis rely on behaviour responses, which are non-existent or very minimal, thereby resulting in misdiagnosis or delayed intervention. Eye tracking provides an objective, non-intrusive means to measure ocular movement, visual interest, and response patterns, which may enable clinicians to make inferences regarding cognitive processing and residual consciousness. Furthermore, these devices provide a window of opportunity for the creation of minimum communication, enabling patients to communicate needs or preferences utilizing gaze-supported interfaces. This paper discusses current eye tracking systems, their application in the clinical field, and the prospect of applying them in the integration of these devices into conventional diagnostic and therapeutic routines. The findings stress that eye tracking not only enhances diagnostic accuracy but offers a platform for patient-centred communication, which eventually contributes to improved clinical outcomes and quality of life. Of the methods discussed (EOG, VOG, and infrared), infrared eye-tracking system had the best overall balance of spatial precision, responsiveness, and patient comfort and thus functioned best for diagnostic detection and communication based on gaze in this population. 2026 IEEE.