Sarifuddin Madenda

Accurate mangrove species identification is fundamental to biodiversity conservation, ecological monitoring, and climate change mitigation because mangrove ecosystems provide essential ecological services and function as significant carbon sinks. This study proposes a systematic optimization framework for mangrove species identification by jointly investigating dataset size, convolutional depth, and hyperparameter configuration within a convolutional neural network (CNN). Leaf images of seven mangrove species collected in Kolaka Regency, Southeast Sulawesi, were augmented, resized to 128 × 128 pixels, and processed using the NVIDIA DGX A100 supercomputer. The framework systematically optimized the number of convolutional layers, learning rate (LR), batch size, number of epochs, kernel size, activation function (AF), and optimizer, followed by a comprehensive evaluation of their combined effects on classification performance. The optimized CNN achieved 98.93% testing accuracy, 98.71% precision, 98.71% recall, and a 98.86% F1-score. The results demonstrate that the integrated optimization of CNN architecture, dataset size, and hyperparameters substantially improves model robustness and generalization over conventional fixed CNN configurations, providing a practical deep learning framework for automated mangrove monitoring and biodiversity assessment.

The object of this study is digital image data consisting of numeric values in the form of pixels. Digital image data become an important in the current era of data and information exchange because the increasing exchange of this data is often vulnerable to attacks or hacking that can cause losses. This study emphasizes the performance of encryption and decryption algorithms using keystreams generated from the composition of two chaos functions. The function is Gauss Iterated and Dyadic Transformation (GIDT). The test results show that the keystream passes 100% of the National Institute of Standards and Technology (NIST) randomness test. The encryption process time speed = 3.92 seconds and the decryption process time speed = 3.97 seconds with City.PNG image test data with dimensions of 512 × 288. The encrypted image test was carried out on the prototype application software resulting in a uniformity test below the critical value = 310.4574, correlation test results = close to zero (0), entropy test results = 7.99, Number of Pixel Change Rate (NPCR) test results = 99.6% and Unified Average Changing Intensity (UACI) test results = 33.45%, key space size = 26.82 × 10651, which shows that the encrypted image is resistant to statistical attacks, differential attacks, and brute-force attacks. The quality testing of the decrypted image was carried out on the prototype application software which produced a Peak Signal-to-Noise Ratio (PSNR) value = ∞, this indicates that the decrypted image is 100% back to the original image. The contribution of this study is to produce a new chaos function as well as encryption and decryption algorithms to significantly speed up processing time without compromising the resilience of image data to attacks

Corneal ulcers can cause permanent vision loss if not diagnosed and managed promptly, particularly in settings with limited access to ophthalmology services. This study aims to develop an automated deep learning approach for classifying corneal ulcer severity from fluorescein slit-lamp images. An EfficientNetV2-S–based model is employed, incorporating corneal area masking to suppress non-relevant regions and class distribution–based augmentation to address data imbalance. To improve evaluation reliability, a leakage-aware data splitting strategy is applied before and after augmentation. Experimental results show that the proposed approach achieves a maximum validation accuracy of 95.93% under non-leakage conditions for the category classification scenario, while maintaining high training efficiency. These results demonstrate that the proposed method provides a robust and efficient solution for automated corneal ulcer severity assessment and has the potential to support clinical decision-making in ophthalmic practice.

This paper presents a mathematical framework for modeling and controlling internet of things (IoT) malware propagation via an adaptive patching strategy governed by the real-time basic reproduction number R₀. We introduce the SEIR-P (susceptible–exposed–infected–recovered–patched) model, where the patching control rate u(t) is a sigmoid feedback function of R₀(t). All epidemiological parameters are calibrated from three empirical malware captures of the IoT-23 dataset (Stratosphere laboratory, Czech Technical University) a publicly available labeled collection of real IoT network traffic comprising 23 captures from infected and benign devices: CTU-IoT-1 (Mirai), CTU-IoT-9 (Torii), and CTU-IoT-17 (IRCBot) yielding the first empirically grounded parameter set for SEIR-type IoT epidemic models with confidence intervals. The optimal control problem is formulated via pontryagin’s maximum principle (PMP), and a closed-form R₀(u) expression is derived via the next-generation matrix (NGM), yielding the critical threshold u_crit = 0.142 day⁻¹. Five comparative simulation scenarios over a 365-day horizon show that the proposed R₀-adaptive strategy achieves a 91.3% reduction in peak infection (360 vs. 4,142 devices), eradicates malware by day 179, and attains the highest cost-effectiveness index (CEI = 1.142). Global asymptotic stability of the disease-free equilibrium under u*(t) is proven via Lyapunov’s method and LaSalle’s Invariance Principle. PRCC sensitivity analysis identifies u_max and β as dominant parameters. This closed-loop framework bridges the gap between abstract epidemic theory and deployable IoT security management.

Mustard greens are an important vegetable commodity, but their production is often affected by pest attacks, especially the cabbage worm Crocidolomia pavonana (C. pavonana). The larvae damage leaf tissues and cause significant yield losses, while chemical control is often ineffective due to differences in insecticide sensitivity across larval instars. This study proposes a deep learning based classification approach combined with gradient weighted class activation mapping (Grad-CAM) to identify larval instars of C. pavonana on mustard plants. A dataset of 684 images covering instars 1 to 4 was collected from laboratory rearing and field observations, then processed using resizing and augmentation techniques and divided into training, validation, and testing sets with an 8 to 1 to 1 ratio. Two convolutional neural network (CNN) models, visual geometry group 19 (VGG19), and Xception, were implemented with additional fully connected layers. The VGG19 model achieved 94.20% accuracy and outperformed Xception. Grad-CAM successfully highlighted larval regions and supported visual interpretation. The results show that the proposed method can improve pest identification accuracy and support more effective pest management.

This study proposes a morphological convolutional neural network (MCNN) architecture that integrates morphological operations with CNN layers for facial expression recognition (FER). Conventional CNN-based FER models primarily rely on appearance features and may be sensitive to illumination and demographic variations. This work investigates whether morphological structural representations provide complementary information to convolutional features. A multi-source and multi-ethnic FER dataset was constructed by combining CK+, JAFFE, KDEF, TFEID, and a newly collected Indonesian Facial Expression dataset, resulting in 3684 images from 326 subjects across seven expression classes. Subject-independent data splitting with 10-fold cross-validation was applied to ensure reliable evaluation. Experimental results show that the proposed MCNN1 model achieves an average accuracy of 88.16%, while the best MCNN2 variant achieves 88.7%, demonstrating competitive performance compared to MobileNetV2 (88.27%), VGG19 (87.58%), and the morphological baseline MNN (50.73%). The proposed model also demonstrates improved computational efficiency, achieving lower inference latency (21%) and reduced GPU memory usage (64%) compared to baseline models. These results indicate that integrating morphological representations into convolutional architectures provides a modest but consistent improvement in FER performance while enhancing generalization and efficiency under heterogeneous data conditions.

Road surface deterioration poses a critical risk to driving safety and comfort, necessitating timely and accurate detection to support effective maintenance. Manual inspection methods are often inefficient, underscoring the need for automated approaches based on computer vision. This study investigates the integration of feature extraction techniques histogram of oriented gradients (HOG) and local binary pattern (LBP) with convolutional neural network (CNN) architectures ResNet50 and InceptionV3 for the classification of road damage. A dataset of 1,580 images was categorized into five damage types: alligator crack, longitudinal crack, other crack, patching, and potholes. Experimental results indicate that HOG–ResNet50 achieved 79% accuracy, while LBP–InceptionV3 yielded the best performance at 97%. The contributions of this study are threefold: i) an automated framework is proposed that combines texture-based features with deep learning for road damage detection, ii) the LBP–InceptionV3 combination is shown to provide superior accuracy compared to conventional pairings, and iii) the approach offers a scalable and reliable alternative to manual inspection methods, supporting more efficient road maintenance planning.

The present paper compares and analyzes the design of AES-128 encryption and decryption using Finite State Machine (FSM) architecture on FPGA and MATLAB platforms. This study aims to evaluate performance disparities in terms of execution time, throughput, and hardware efficiency under identical input data and key conditions. The FSM-based AES algorithm was modeled in MATLAB for functional validation and synthesized on an Artix-7 FPGA using VHDL. The experimental results confirmed that both platforms produced identical ciphertext and plaintext outputs, verifying the correctness of the processes employed. However, the FPGA demonstrated significantly better performance in terms of execution speed. Encryption and decryption times were measured in microseconds on the FPGA, while similar operations on the MATLAB platform required hundreds of milliseconds. The FPGA implementation achieved throughput of 872.53 Mbps for encryption and 858.49 Mbps for decryption with area usage of 1263 and 1428 slices, respectively. This yields an efficiency of 0.691 and 0.601 Mbps/slice, which is considered efficient according to established benchmarks. Compared to previous MATLAB-only and FPGA pipelined implementations, the current design strikes a balance between resource usage and performance, making it ideal for lightweight cryptographic applications in embedded systems. These results provide practical insights into selecting platforms for secure, real-time data processing.

The security of digital image data has become increasingly critical in modern communication systems. While chaos-based cryptography offers a promising solution, many existing algorithms lack rigorous security validation. This paper introduces the Logistic-Circle Map (LC Map), a novel one-dimensional compound chaotic system designed to provide a robust and efficient foundation for image encryption. By composing the Logistic Map and the Circle Map, the LC Map exhibits a broader chaotic range and higher dynamical complexity. The performance and security of an LC Map-based encryption scheme are extensively validated using a comprehensive dataset of 24 digital images. Security analysis demonstrates that the algorithm is highly resistant to brute-force, statistical, and differential attacks. It provides a vast key space and demonstrates very strong key sensitivity, both confirmed through experimental evaluation. Test results show near-ideal performance on standard security metrics, with a Number of Pixels Change Rate (NPCR) approaching 99.6%, a Unified Average Changing Intensity (UACI) approaching 33.4%, and an information entropy value nearing the theoretical maximum of 8. Further quantitative comparative analysis demonstrates the superiority of the LC Map in balancing security and computational efficiency. Thus, the LC Map is presented as a rigorously validated component for the development of future image cryptosystems.

One effort to maintain the validity of the Al-Qur'an manuscript is the analysis and verification of the manuscript by experts (Pentashih). Currently, manuscript verification without translation takes 30 working days. Therefore, to support Pentashih in reviewing the manuscript, technology is needed to expedite the Pentashih process and prevent analysis errors caused by Pentashih fatigue. This study conducts a writing analysis of the target manuscript by referring to the template manuscript, implementing image preprocessing stages, applying SSIM for analysis, and employing the pixel-matching method. This method examines the manuscript's writing by comparing two block images at the pixel level. Block images are produced by preprocessing the manuscript images before image-matching analysis is performed. Image preprocessing comprises: cropping the outer frame, cropping the inner frame, segmenting the page into row images, adjusting margins, aligning image sizes, segmenting rows into block images, and aligning positions between block images. Pixel value differences are calculated at the same positions across each column and row of the template and target block images. Block image positions with pixel values ≥ 200 occur in 5 consecutive columns, adjacent rows with a distance = 1, and an SSIM value ≥ 0.9, both images meet the mismatch criteria. These findings indicate that the proposed approach provides an efficient and accurate solution for automating the verification of the Indonesian Standard Mushaf.