Amal Ajayan

Diseases that affect plant leaves have a big effect on food supply and agricultural output. This makes it very important to find accurate and automatic ways to diagnose these diseases. This paper presents a segmentation-aware hybrid deep learning (DL) approach that integrates both Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) for the effective classification of plant leaf diseases. An unsupervised K-means clustering method is used first to make rough segmentation masks that don’t need manual labels. Then, a U-Net model is used to make them better. The combination of CNN and ViT is used to get at the segmented leaf images and get both main disease-related features and a broader context. Gradient-weighted Class Activation Mapping is used to show which parts of the leaf affect the diagnosis. This makes it easier to understand the model’s decisions. We test the framework on the PlantVillage dataset to see how well it works and then again on the Plant Disease Detection dataset to see how it works with different types of this data. The results show that the disease classification is better, that the system works better when the data sources change, and that the visual insights are very clear. These results show that the suggested approach is a dependable and comprehensible solution for the automated plant disease identification and can facilitate precision agriculture.

The Internet of Medical Things (IoMT) increasingly relies on continuous sensing and cloud-assisted diagnostics, yet its devices operate under stringent resource constraints that limit the deployment of conventional cryptographic mechanisms. To address this challenge, we propose a lightweight encryption framework that integrates the National Institute of Standards and Technology (NIST)-standardized ASCON-v1.2 authenticated encryption scheme with two optimization strategies: Hypercube Optimal Search (HOS) and Modified Wild Geese (MWG). The framework systematically explores ASCON’s implementation-level parameter space and identifies configurations that preserve security margins while reducing computational cost, latency, and energy consumption on constrained IoMT hardware. HOS performs multidimensional search and contraction over feasible parameter regions, whereas MWG provides adaptive local refinement to optimise performance under varying device conditions. An experimental evaluation of publicly available medical image datasets demonstrates that the proposed HOS + MWG + ASCON pipeline enhances diffusion and randomness characteristics while reducing execution time compared to baseline ASCON configurations. Quantitative assessments using Peak Signal-to-Noise Ratio (PSNR), Mean Squared Error (MSE), Structural Similarity Index Measure (SSIM), Number of Pixels Change Rate (NPCR), Unified Average Changing Intensity (UACI), entropy, correlation coefficients (CC), and Bit Error Rate (BER) confirm the robustness of the system against statistical, differential, and structural inference attacks. Hardware tests on an ESP32-S3 platform further validate the framework’s energy efficiency and reliability under realistic IoMT constraints. Overall, this work presents a scalable, computationally efficient, and security-preserving encryption solution tailored for real-time IoMT deployments, where confidentiality, latency, and energy efficiency are all critical.

KSII Transactions on Internet and Information Systems TIIS Paper Details

Android ransomware has emerged as a major threat to mobile ecosystems, leveraging obfuscated payloads and dynamic command-and-control channels to evade conventional detection systems. Existing approaches often rely on static, batch-trained models that lack adaptability to evolving threat behaviors, resulting in degraded accuracy over time due to concept drift. This presents a critical challenge for real-time deployment, as new ransomware variants continually mutate their signatures and alter network traffic patterns to evade detection. To bridge this gap, this study proposes a robust ensemble-based machine learning framework for proactive detection of Android ransomware using network traffic metadata. The framework integrates advanced classifiers, including Light Gradient Boosting Machine, eXtreme Gradient Boosting Machine, and Random Forest, with Synthetic Minority Oversampling Technique enhanced stratified cross-validation to mitigate class imbalance and improve generalizability. Furthermore, explainable artificial intelligence methods such as SHapley Additive exPlanations and Local Interpretable Model-Agnostic Explanations are employed to enhance interpretability and analyst trust. In the context of ransomware detection, the importance of online learning lies in its ability to adapt to evolving threat patterns in real time. Ransomware frequently mutates payload signatures and obfuscates behavioral traces, causing traditional models to deteriorate under changing data distributions. To address this, we conducted a concept drift evaluation using an incremental LightGBM model, tested on chronologically partitioned traffic data across five temporal blocks. This approach enables continuous adaptation to new data streams without requiring full retraining, thereby maintaining detection robustness and reducing false negatives in production. Experimental results on a balanced dataset demonstrate that LightGBM achieves the highest classification performance, indicating the efficacy and adaptability of the proposed framework for real-time Android ransomware mitigation in dynamic network environments.

The exponential expansion of the Internet of Things (IoT), Industrial Internet of Things (IIoT), and Transportation Management of Things (TMoT) produces vast amounts of real-time streaming data. Ensuring system dependability,... | Find, read and cite all the research you need on Tech Science Press