Journal of Computers and Intelligent Systems

Detection and Classification of Alopecia Areata Using Diverse Feature-Set

Amna Noreen, Haider Khan, Syed M Adnan Shah, Wakeel Ahmed, Syed Haseeb Amjad — Fri, 27 Jun 2025 00:00:00 +0000

Alopecia areata is a prevalent autoimmune disorder resulting in hair loss on the scalp and other body parts, affecting millions worldwide. This condition can significantly impact one's psychological well-being and self-esteem, highlighting the importance of early detection for better disease management and potential hair regrowth. This study uses computer vision techniques to propose a comprehensive method for detecting alopecia areata from camera images. Two distinct datasets, Dermnet for alopecia areata images and Figaro1k for healthy images are employed, with a preprocessing phase involving histogram equalization to enhance image quality. Subsequently, color, texture, and shape features are extracted from the images, followed by feature fusion and selection to maximize the discriminative power of the dataset. Several popular classifiers, such as Random Forest, SVM, Decision Tree, Logistic Regression, Naive Bayes, ANN, and KNN, are employed to identify the most effective models. The proposed technique achieves a remarkable accuracy of 96.43%, outperforming related research methods. This study shows considerable promise for improving the early detection and treatment of Alopecia Areata by utilizing sophisticated computer vision and machine learning methods, leading to enhanced patient outcomes and a better quality of life.

Data Enhancing Cybersecurity through Botnet Security Isolation for Smart IoT Devices: A Deep Learning Approach

Fri, 27 Jun 2025 00:00:00 +0000

The Internet of Things (IoT) sector continues to expand rapidly to connect more than billions of devices throughout healthcare settings as well as transportation domains and smart residential spaces. Amazing network connectivity affords IoT systems to multiple security problems especially through botnet attacks that utilize illegally gained control over IoT devices to carry out harmful operations. Security solutions from the past struggle to stop and address these attacks because IoT devices present various resource limitations together with their diverse operational characteristics. The proposed research presents a deep learning botnet detection system for IoT networks by applying Convolutional Neural Networks (CNNs) together with Long Short-Term Memory Networks (LSTMs) and Autoencoders for analyzing IoT traffic patterns. The models received training using Bot-IoT dataset to find their optimal performance through accuracy and precision and recall and F1-score evaluations. CNN generates better results than other examined models by achieving 94% accuracy while LSTM obtains 92% and Autoencoder provides 88%. The research established CNN as the best model for traditional botnet detection yet LSTM showed exceptional capability in detecting temporal patterns and Autoencoder achieved the best results for identifying new botnet traffic types. The system displays encouraging performance however it encounters technical challenges because of its issues with processing real-time data and model scalability issues as well as unbalanced IoT datasets. According to research findings deep learning models specifically Convolutional Neural Networks demonstrate substantial potential for enhancing botnet detection but ongoing research must focus on performance enhancement techniques for realistic ecosystem deployment and handling various IoT network configurations and scalability considerations.

GoogleNet’s Image Classification Performance Analysis: Effects of Dataset Size, Balancing, and Splits Ratios

Mujeeb Ur Rehman, Mariyam Amreen — Fri, 27 Jun 2025 00:00:00 +0000

In computer vision, image categorizing plays a condemnatory role in many real-world applications. GoogleNet is a intimate deep learning model especially expand for image clas- sification and object recognition tasks. It effectually categorizes images based on learned patterns.This study evaluates the per- formance of the GoogleNet architecture in image classification, focusing on three key aspects: the impact of dataset size, dataset balancing, and different train-test split ratios. We examined the model’s accuracy on the CIFAR-10 dataset by varying dataset sizes at 25%, 50%, 75%, and 100%, considering both balanced and unbalanced data. The findings indicate that dataset size and balance notably affect classification accuracy, with balanced datasets generally outperforming unbalanced ones. Additionally, when experimenting with different train-test split ratios 50%- 50%, 60%-40%, 70%-30%, 80%-20%, and 90%-10% the model achieved the best results when trained on 70% of the data and tested on the remaining 30%.

Cyber Threat and Vulnerability Classification Using NLP and Machine Learning Techniques on Text-Based Security Data

Talha Khan, Mubasher Malik, Zahid Aziz, Muhammad Kamran Abid, Muhammad Sabir — Fri, 27 Jun 2025 00:00:00 +0000

The rapidly developing cybersecurity sector faces the essential problem of detecting and classifying cyber threats with precision. The rise of complicated data and its growing volume requires machine learning (ML) techniques to successfully automate threat detection operations through modern methods. The research evaluates six different ML algorithms for cybersecurity threat classification through Logistic Regression, SVM, Random Forest, Naive Bayes, LSTM, and BERT performance analysis. The systematic evaluation methodology analyzes these models by measuring their accuracy, together with precision and recall metrics, along with F1-score and execution time efficiency. Our examination starts with tokenization, then carries out stop-word elimination before performing TF-IDF vectorization for model enhancement purposes through various feature encoding approaches. The study examines the effects that employing both categorical and continuous feature encoding methods has on the outcomes. The research makes its original contribution through analyzing performance-speed tradeoffs between deep learning models and standard models applied to cybersecurity contexts. BERT proves to be the superior model since it delivers 93.8% accuracy and 96.2% ROC-AUC score at the cost of increased computational requirements. Random Forest and SVM exhibited comparable results, but Naive Bayes demonstrated the least effective performance with accuracy and recall statistics. BERT outperforms other models in cybersecurity, but its high computing requirements prevent it from real-time implementation.

Applications of Stochastic Processes in Quantum cryptography for Secure Cryptographic Protocols

Fri, 27 Jun 2025 00:00:00 +0000

Quantum cryptography offers higher security than traditional cryptography but faces challenges due to noise and interference. Stochastic processes, particularly randomness and uncertainty, provide tools to model and protect quantum systems. This review paper explores the role of stochastic processes in quantum cryptography, focusing on quantum entanglement and secure protocols like Quantum Key Distribution (QKD). Key stochastic concepts, including probability distributions, Markov processes, and noise effects in quantum mechanics, are introduced. The research examines how stochastic models enhance performance and security in entanglement-based protocols while addressing challenges such as noise, decoherence, scalability, hardware limitations, and attack vulnerabilities. Potential research directions include efficient quantum error correction, quantum networking, and integrating classical and quantum cryptography to improve security and practicality