Zaffar Ahmed Shaikh

This paper presents a comprehensive framework for dynamic soft sets and dynamic hypersoft sets, along with their associated distance and similarity measures. The study begins with classical soft and hypersoft sets, highlights their limitations in handling temporal variation and evolving parameter structures, and then extends them to dynamic settings suitable for real-world decision-making problems. Formal definitions, theoretical properties, and illustrative examples are provided to show how the proposed framework can model time-dependent situations such as healthcare monitoring, financial risk assessment, and multi-attribute diagnosis. In addition, multiple distance measures and their corresponding similarity measures are introduced for Dynamic Single-Valued Neutrosophic Hypersoft Sets (DSVNHSS), enabling flexible comparison of alternatives under uncertainty. A case study on ranking cities by air quality and livability demonstrates how the proposed similarity-based approach can support practical multi-criteria decision-making. The results show that DSVNHSS offer a richer and more adaptive modeling environment than classical soft and hypersoft sets, making them highly suitable for applications where attributes, relevance, and system states evolve over time.

Latency-sensitive and context-driven resource orchestration is needed to maintain service continuity in human-interactive ubiquitous settings. Agile Context-Adaptable Interaction Scheme (ACAIS) is a unified interaction–context–resource co-optimization system that treats resource management as a multi-objective adaptive decision process. ACAIS dynamically allocates computing resources under restricted capacity restrictions using a deep reinforcement learning–based policy optimizer and a structured state-space representation of contextual entropy, interaction intensity, and workload elasticity. In real time, a dual-mode latency inference system examines context-to-latency and interaction-to-latency mappings to manage allocation strategies. An external benchmark dataset shows significant performance gains under peak workload conditions (12 interactions/hour), including 12.68% latency reduction (p < 0.01), 12.85% adaptability improvement, 8.01% resource utilization efficiency enhancement, and 15.54% context-awareness accuracy increase compared to baseline adaptive schemes. ACAIS' predictive, interaction-aware resource governance paradigm for scaled ubiquitous computing environments is confirmed by these findings.

Consumer health applications for diabetes management create critical privacy vulnerabilities when sharing sensitive patient data across healthcare ecosystems, demanding lightweight cryptographic solutions for resource-constrained mobile environments. This paper presents NeuroShield-Diabetes, a hybrid framework integrating neural cryptography with machine learning for privacy-preserving diabetes data sharing on smartphones and wearables. The framework employs tree parity machines for neural key exchange, achieving cryptographic strength comparable to 2048-bit RSA with 91% reduced computational complexity measured as CPU cycles per key generation operation. Adaptive differential privacy with calibrated noise injection maintains formal guarantees (ε ≤ 0.8 for high-risk queries) while preserving 94.7% data utility for glycemic prediction. Evaluation across 5,600 diabetes patients using MIMIC-III and UCI Diabetes datasets demonstrates 73% reduction in data sharing latency versus conventional cryptographic solutions. The neural intrusion detection system achieves 92.6% accuracy in identifying malicious access attempts with sub-120ms overhead. Field deployment confirms HIPAA compliance and GDPR conformance while maintaining interoperability with existing healthcare infrastructures, establishing NeuroShield-Diabetes as a practical privacy-preserving solution for consumer diabetes applications.

This study introduces the structure of normalized Triangular Type-2 Intuitionistic Fuzzy Numbers (TrT2IFNs), a relatively recent concept with strong potential to enhance the modeling and analysis of uncertainty. The study also presents basic arithmetic operations and ranking functions for these fuzzy parameters. The arithmetic operations are defined using the (α, β)-cut technique, where fuzzy numbers are decomposed into α- and β-level sets to enable systematic computation. In addition, a new ranking function is developed based on a mean interval approach, enabling effective comparison of these fuzzy numbers. To validate the proposed methods, a Transportation Problem (TP) is formulated, where cost parameters are represented by TrT2IFNs. The model is illustrated through numerical examples, and the results are analyzed. The findings show that the proposed approach addresses uncertainty in transportation problems more effectively than existing methods. Applying TrT2IFNs improves the representation of cost uncertainty, supporting better decision-making in logistics and supply chain optimization. The numerical results indicate reduced transportation costs compared to current approaches, demonstrating the practical value of the proposed framework in operations research under real-world uncertainty.

The proliferation of IoT devices has expanded the attack surface, making zero-day threat detection an urgent priority. This paper proposes LLMEdgeSec, a framework that leverages large language models (LLMs) for log reasoning to detect and contextualize zero-day exploits in real-time. Unlike conventional anomaly detectors that rely on static signatures or pre-defined rules, LLMEdgeSec applies prompt-based reasoning over system logs, communication metadata, and device telemetry, enabling proactive identification of previously unseen attack vectors. The framework employs a dual-stream architecture: (1) a lightweight transformer encoder on edge nodes for local feature extraction, and (2) a cloud-based LLM interface for contextual reasoning and cross-device threat correlation. A zero-shot calibration method mitigates hallucinations and reduces false positives. Evaluation on TON IoT, CIC-IDS2018, and a real-world smart factory dataset demonstrates F1 scores exceeding 92% on zero-day samples with communication overhead below 8MB per federated round. Results show significant improvements in accuracy, interpretability, and adaptability over state-of-the-art baselines.

Smart health devices are playing an increasingly vital role in early diagnosis and personalized care, especially for complex conditions like thyroid cancer. Yet, making sense of subtle, time-dependent patterns in patient data remains challenging— particularly when diagnostic accuracy, transparency, and real-time responsiveness all matter. To tackle this, we introduce the Attention-driven Quantum-Explainable Unified Block with integrated Diffusion Modeling (ThyroDiff-X). ThyroDiff-X brings together three complementary components to model evolving clinical trends. This integration enables the model to capture longitudinal biomarker evolution and suppress noise through diffusion-based temporal reconstruction, explaining its strong performance gains over static and single-visit diagnostic approaches. Specifically, quantum-inspired sinusoidal encoding enriches feature representation, a context-aware diffusion module (Enhanced-DDIM) reconstructs temporal biomarker trajectories using visit-level memory and SHAP-informed clinical context, and multi-head attention adaptively focuses on the most relevant visits. A built-in SHAP layer adds interpretability by highlighting the key features behind each prediction. To optimize performance, the model leverages MSAMO-VCS—a hybrid tuning approach that combines wide-ranging parameter exploration with precise local refinement. The model was rigorously tested across four diverse datasets—SmartThyro-300, UCI Thyroid, Sick-Euthyroid, and THYROID-MVTS—where it consistently outperformed leading baselines, achieving up to 99.5% accuracy, a log loss as low as 0.04, and the highest Quantum Explainability Confidence Rate (QXCR). Its successful deployment on a Jetson Nano board confirmed its real-time readiness and efficiency under resource-constrained settings.

Pakistan ranks among the top ten nations globally that are likely to experience severe impacts from climate change and are facing critical food security challenges. Variability in temperature and rainfall alters the geographical suitability of land for crop cultivation. This study employed the Maximum Entropy (MaxEnt) model to forecast the effects of climate change and evaluate land suitability for Triticum aestivum L. (common wheat) production in Pakistan, where it is a major staple crop. Bioclimatic variables and occurrence data for two climatic scenarios, Representative Concentration Pathways (RCP) 4.5 and 8.5, from five General Circulation Models (GCMs) were used for the year 2070. The prime factors affecting the Triticum aestivum L. distribution are temperature seasonality, annual precipitation, and mean temperature of the warmest quarter. The findings indicate an average decline in highly suitable and moderately suitable areas, whereas an increase is observed in the least suitable areas in future scenarios. The highly suitable area for future distribution accounts for 26.78% and 19.67% under RCP 4.5 and RCP 8.5, respectively, highlighting a negative effect on prospective wheat production in Pakistan. The outcome of this research is of utmost significance for decision-makers to develop suitable adaptation and mitigation protocols to sustain wheat productivity under a changing climate.

The growing reliance on Internet of Things (IoT) networks in marine environments has highlighted concerns about sensor spoofing attacks, particularly under the disruptive influence of harmful algal blooms (HABs). These blooms introduce considerable biological noise and sensor drift, creating a challenging environment where distinguishing between genuine environmental anomalies and deliberate adversarial attacks becomes difficult. This study tackles the pressing issue of spoofing detection in marine IoT systems affected by bloom-induced interference, aiming to offer a solution that is both accurate and interpretable. The proposed TransformerXL-GAN with explainable artificial intelligence (XAI) framework combines TransformerXL’s long-term temporal modeling with GAN’s adversarial robustness while incorporating SHapley Additive exPlanations (SHAP)-based interpretability for clearer decision-making insights. Additional strategies, such as event-aware data augmentation, temporal anomaly filtering, and hybrid Cascaded Hyperparameter–Multi-Stage Adaptation (CH-MSA) optimization, further bolster learning stability and detection precision. Validation is conducted using the MARSS-2025 dataset, a comprehensive collection of marine sensor readings from the Norwegian Institute for Water Research. Experimental outcomes reveal that the TransformerXL-GAN+XAI framework surpasses both traditional and cutting-edge deep learning methods, achieving an impressive 98.8% accuracy and an AUC of 0.995. An accompanying ablation study underscores the collective benefit of each architectural component. These findings demonstrate the framework’s potential to deliver reliable, transparent, and biologically adaptive spoofing detection in complex marine IoT environments.