Mohiuddin Mehedi

The exponential growth of interconnected systems has dramatically expanded the attack surface of modern networks, making zero-day attacks, which exploit previously unknown vulnerabilities, a critical threat to organizational security. Traditional signature-based and anomaly detection systems often struggle to identify such previously unseen attack patterns, and while deep learning (DL) offers promising detection capabilities, existing DL-based intrusion detection systems (IDS) suffer from poor generalization to novel attacks and operate as opaque black boxes, lacking interpretability for security analysts. To address these limitations, this paper proposes ZD-HybridNet (Zero-Day Hybrid Attention Network), a novel explainable deep learning framework for large-scale network traffic analysis. ZD-HybridNet integrates a Transformer encoder branch for capturing global feature dependencies and an LSTM branch for modeling sequential traffic patterns, dynamically fused via an adaptive attention-based gating mechanism. This architecture is coupled with an integrated explainability framework that combines model-internal attention and fusion signals with SHAP-based feature attribution to support interpretation of detection decisions. Evaluated on the UNSW-NB15 dataset under a controlled category-holdout protocol in which Shellcode and Worms are entirely excluded from training and validation, ZD-HybridNet achieves an accuracy of 94.04%, a precision of 96.76%, a recall of 93.96%, and an F1-score of 95.34%, outperforming standalone LSTM, Transformer, FT-Transformer, and 1D CNN baselines under the same experimental protocol. Precision-recall, ROC, zero-day-specific, ablation, and SHAP-based analyses further support the model’s detection performance and interpretability within the defined Shellcode–Worms holdout setting, while deployment-oriented computational validation remains necessary.

Despite the benefits associated with the use of smart meters in advanced metering infrastructure, the widespread deployment of such meters has introduced vulnerabilities that leave power systems susceptible to stealthy false data injection attacks which cannot be detected by conventional methods. In this paper, we present a framework that combines a bidirectional long short-term memory network with an attention mechanism and a blockchain integrity layer to provide secure anomaly detection. We utilize the Smart Meter Electricity Consumption Dataset, augmented with synthetically injected anomalies, to detect abnormal consumption behavior. Experimental results demonstrate that the proposed hybrid model achieves up to 96.23% accuracy, 99.88% precision, 92.58% recall, and 96.09% F1-score, outperforming eXtreme Gradient Boosting (XGBoost), Isolation Forest, and Random Forest baselines. Confusion matrix analysis confirms minimal false positives and strong detection capability for stealthy attacks. The blockchain layer ensures immutability and trustworthiness of detection results through cryptographic hashing and consensus mechanisms with negligible overhead. The proposed framework offers a scalable, secure, and interpretable solution for defending smart grid infrastructures against complex cyberattacks.

Customer churn is a critical challenge for subscription-based digital platforms, as customer activity patterns change dynamically over time. Traditional churn modeling methods fail to account for sequential dependencies across customer interaction histories. This paper presents Attn-ChurnNet, a novel attention-based Transformer architecture that effectively predicts customer churn by modeling sequential customer contact histories. The proposed methodology leverages multi-head self-attention with sinusoidal positional encodings and Pre-LN residual connections to highlight key interaction sequences and interpret the temporal dynamics of customer activity in subscription platforms. Experiments are conducted on the large-scale WSDM–KKBox Customer Churn Prediction dataset using a temporal train/validate/test split, incorporating transaction records, usage logs, and customer demographic information. Comprehensive comparison against established baselines—Logistic Regression (LR), Random Forest (RF), XGBoost, and Gated Recurrent Unit (GRU)—demonstrates that Attn-ChurnNet achieves a macro-averaged classification accuracy of 97.03% (±0.41%), precision of 95% (±0.5%), recall of 96% (±0.6%), F1-score of 95.50% (±0.5%), AUC of 0.98 (±0.004), Average Precision of 0.963, and log-loss of 0.15 (±0.007) under five-fold stratified cross-validation, outperforming all competing approaches with statistical significance (𝑝<0.01, McNemar’s test). A comprehensive two-part ablation study (22 variants), calibration analysis (ECE = 0.031; 𝑇∗𝑠 = 1.08), attention entropy analysis with Jensen–Shannon divergence and two-sample t-test (𝑡=18.4, 𝑝<0.001), Integrated Gradients attribution, conformal prediction (91.4% coverage, 88% singleton efficiency), precision–recall analysis, and computational complexity evaluation further validate the model’s design and production readiness.

Alzheimer's Disease (AD) constitutes a progressive neurodegenerative pathology that undermines cognitive faculties, mnemonic capacity, and the aptitude for executing quotidian tasks in geriatric individuals, frequently accompanied by alterations in demeanor and temperament. In the absence of a definitive cure, extant therapeutic interventions demonstrate optimal efficacy during the initial and intermediary phases of AD, underscoring the criticality of precocious diagnosis amidst the burgeoning geriatric demographic and the concomitant escalation in AD prevalence. This inquiry endeavors to discern incipient indicators of AD by harnessing machine learning and deep learning methodologies to scrutinize MRI imagery derived from the Open Access Series of Imaging Studies (OASIS) dataset. A spectrum of algorithms was deployed, encompassing Random Forest and Logistic Regression within traditional machine learning paradigms, Extra Trees as an exemplar of ensemble learning, and Convolutional Neural Networks (CNN) representative of deep learning architectures. The efficacy of these models was appraised through metrics including accuracy, precision, recall, and area under the curve (AUC), wherein CNN exhibited preeminent accuracy and AUC metrics, while Extra Trees distinguished itself in precision and recall, substantiating the capacity of both deep learning and ensemble learning modalities to deliver robust performance in the early detection of AD.

Cardiovascular diseases (CVDs) remain the leading cause of death globally, creating a persistent need for early screening tools that are both accurate and clinically interpretable.1 This study presents an end-to-end, explainable machine learning framework for binary heart disease prediction using a structured clinical dataset derived from the UCI Heart Disease benchmark, where published experiments commonly use a 14-variable subset and focus on distinguishing presence versus absence of disease.2 To strengthen reliability and reduce bias, the pipeline integrates stratified train–test splitting, feature scaling for scale-sensitive learners, and Synthetic Minority Over-sampling Technique (SMOTE) to address potential class imbalance in the training split.3 Multiple models are compared, including Logistic Regression, Naïve Bayes, KNN, SVM, Decision Tree, Random Forest, Gradient Boosting, AdaBoost, Extra Trees, XGBoost, and MLP. Performance is evaluated using accuracy, precision, recall, F1-score, and ROC-AUC, with 5-fold cross-validation to estimate generalization stability. On the held-out test set, Extra Trees achieved the highest ROC-AUC (90.80%), while SVM obtained the highest accuracy (83.61%). Cross-validation ranked Random Forest (mean ROC-AUC ≈ 90.06%) and AdaBoost (≈ 89.95%) as top performers, and GridSearchCV further optimized Extra Trees to a best cross-validated ROC-AUC of 0.912. Finally, explainability is provided through SHAP, which attributes predictions to clinically meaningful features, supporting transparent decision support rather than black-box output.4