ملف المستخدم
صورة الملف الشخصي

فرج معاذ جاسم

إرسال رسالة

التخصص: علوم حاسوب

الجامعة: جامعة الانبار

النقاط:

17.5
معامل الإنتاج البحثي

الخبرات العلمية

  • تقييم ومراجعة البحوث
  • كتابة البحوث

الأبحاث المنشورة

Enhancing The Security Of The Bitcoin Wallet Master Seed

المجلة: ausrevista

سنة النشر: 2019

تاريخ النشر: 2019-01-09

0

Enhancement of digital signature algorithm in bitcoin wallet

المجلة: Bulletin of Electrical Engineering and Informatics

سنة النشر: 2022

تاريخ النشر: 2021-02-01

Bitcoin is a peer-to-peer cryptocurrency popular for its anonymity, privacy, and low transaction costs, though its wallets face security challenges. This paper proposes a secure key management system with an image-based passphrase for cold wallets to enhance entropy and a modified key generation algorithm for hot wallets that generates a fresh key pair for each transaction. The findings show improved resistance to dictionary attacks, enhanced anonymity, and privacy, with a transaction signing time of ~70 milliseconds.

Finger vein and hand-dorsal vein multimodal biometric system based on convolution neural network

المجلة: AIP Conference Proceedings 2400

سنة النشر: 2022

تاريخ النشر: 2022-10-31

The rise in cybercrime has driven the need for more accurate biometric systems. This paper combines finger vein and hand-dorsal vein traits using a CNN model with an expansion technique to prevent overfitting. Tested on two datasets, the system achieved 96% and 97% accuracy with different fusion methods, reaching 100% accuracy with expansion.

Eurasian oystercatcher optimiser: New meta-heuristic algorithm

المجلة: journal of Intelligent Systems

سنة النشر: 2023

تاريخ النشر: 2022-02-09

This study introduces the Eurasian oystercatcher optimiser (EOO), a meta-heuristic inspired by the bird's food search behavior. Tested on 58 functions, EOO outperforms several well-known algorithms by achieving a good balance between exploration and exploitation while avoiding local optima.

Classifying cuneiform symbols using machine learning algorithms with unigram features on a balanced dataset

المجلة: journal of Intelligent Systems

سنة النشر: 2023

تاريخ النشر: 2023-09-25

he study addresses class imbalance in the Cuneiform Language Identification (CLI) dataset by using oversampling and evaluating various machine learning algorithms. Results show improved performance with accuracies up to 95.46% for traditional models and 93% for deep neural networks.

Extensive Review of State-of-the-Art Classification Techniques for Cuneiform Symbol Imaging: Open Issues and Challenges

المجلة: Iraqi Journal for Computer Science and Mathematics

سنة النشر: 2023

تاريخ النشر: 2023-08-03

This paper explores the challenges of reading ancient cuneiform tablets and the need for digitization to advance research. It discusses the use of artificial intelligence for categorizing and analyzing cuneiform texts, evaluates various methods, and suggests future research directions.

Image Steganography Technique based on Lorenz Chaotic System and Bloom Filter

المجلة: International Journal of Computing and Digital Systems

سنة النشر: 2024

تاريخ النشر: 2024-08-10

The paper introduces a steganography technique using chaotic systems to enhance data concealment in images. By employing Lorenz's chaotic system and Bloom filters, it embeds encrypted data at random positions, overcoming the weaknesses of traditional Least Significant Bit (LSB) methods. The approach improves security against detection and steganalysis, achieving a PSNR of 48.57% and an NPCR of 32.35%.