Attribute-Based Publishing with Hidden Credentials and Hidden Policies Using Group-Signature Authentication Mechanism
Embedded Files
This project investigates and enhances the Privacy Enhanced Attribute-based Publishing of Data (PEAPOD) system. It introduces a group signature-based mechanism to address authentication issues while preserving hidden credentials and policies. By implementing RSA and ElGamal group signature schemes, the work ensures secure sender authentication, prevents inference attacks, and proposes optimizations to improve system efficiency and robustness. This research opens new directions for privacy-preserving attribute-based data sharing systems.
Github | Code | Paper
A Survey on the Application of Generative Adversarial Networks in Cybersecurity: Prospective, Direction and Open Research Scopes
This project surveys the applications of Generative Adversarial Networks (GANs) in cybersecurity. It examines their role in addressing challenges like malware detection, intrusion detection, and adversarial attacks. By analyzing state-of-the-art research, the study highlights the potential of GANs in enhancing defenses, discusses limitations, and proposes future research directions. Key contributions include a novel taxonomy, comparative analyses, and practical insights into GAN-based cybersecurity solutions.
Integrating APK Image and Text Data for Enhanced Threat Detection: A Multimodal Deep Learning Approach to Android Malware (Ongoing)
In this project, we developed a multimodal deep learning framework to enhance Android malware detection by integrating image and textual data from APK files. Using state-of-the-art CNN architectures like ResNet-152 and EfficientNet-B4, I systematically evaluated grayscale and RGB image representations across various resolutions, identifying high-resolution RGB images (256x256, 512x512) as optimal for malware classification with accuracy rates up to 96%. Additionally, we incorporated text data using the CLIP model and the LLaMA-2 language model to analyze permissions and metadata, creating a comprehensive detection system. While the multimodal approach revealed potential, challenges with small datasets highlighted opportunities for improvement through advanced data fusion techniques and larger datasets. This project demonstrates my expertise in machine learning, data processing, and cybersecurity innovation.
Exploring BERT-Based Cross-Lingual Models for Identifying Cyberbullies' Underlying Personality Traits
This project investigates the use of multilingual and cross-lingual BERT-based models to identify the underlying personality traits of cyberbullies across diverse languages. The framework developed combines data preprocessing, model fine-tuning, and performance evaluation using English, Bangla, Hindi, and German datasets. Key findings highlight the efficacy of multilingual BERT (mBERT) and cultural influences on personality trait expression in cyberbullying contexts. The study paves the way for personalized intervention strategies and advancements in multilingual machine learning applications.
Survey and Analysis of IoT Operating Systems: A Comparative Study on the Effectiveness and Acquisition Time of Open Source Digital Forensics Tools
This project investigates the effectiveness and efficiency of open-source digital forensic tools in analyzing data from diverse Internet-of-Things (IoT) devices. By configuring multiple IoT operating systems and conducting standard forensic tasks, the study provides actionable insights into tool performance, computational cost, and data integrity. The findings aim to guide IoT security experts and forensic practitioners in selecting optimal tools for reliable and efficient digital investigations.
Page updated
Google Sites
Report abuse