An Abnormal External Link Detection Algorithm Based on Multi-Modal Fusion

International Journal of Information Security and Privacy, Journal Year: 2024, Volume and Issue: 18(1), P. 1 - 15

Published: Feb. 7, 2024

Website link detection is an important means to ensure the security of external chain. In past, it was mainly realized through blacklisting and feature engineering-based machine learning, which has problems slow speed weak model generalization ability. The development neural networks brought a new solution chain website. To address performance bottleneck caused by variable content length web pages, this article introduces innovative approach: website algorithm based on multi-modal fusion. It extracts text, dynamic script, image features separately, constructs deep fusion that combines these features. Compared with previous research results, proposed method superior traditional single-mode method, can quickly accurately identify malicious pages. accuracy F1 value are improved 2.7% 0.026.

Language: Английский

---

Deep Learning-Based Speech and Vision Synthesis to Improve Phishing Attack Detection through a Multi-layer Adaptive Framework

Published: Feb. 27, 2024

The ever-evolving ways attacker continues to im prove their phishing techniques bypass existing state-of-the-art detection methods pose a mountain of challenges researchers in both industry and academia research due the inability current approaches detect complex attack. Thus, anti-phishing remain vulnerable because increasingly sophistication tactics adopted by coupled with rate at which new are being developed evade detection. In this research, we proposed an adaptable framework that combines Deep learning Randon Forest read images, synthesize speech from deep-fake videos, natural language processing various predictions layered significantly increase performance machine models for attack

Language: Английский

---

An Investigation into the Performances of the State-of-the-art Machine Learning Approaches for Various Cyber-attack Detection: A Survey

Published: Feb. 27, 2024

To secure computers and information systems from attackers taking advantage of vulnerabilities in the system to commit cybercrime, several methods have been proposed for real-time detection improve security around systems. Of all methods, machine learning had most effective method securing a with capabilities ranging early software ongoing compromise system. As there are different types cyberattacks, each existing state-of-the-art models depends on algorithms training which also impact their suitability particular type cyberattack. In this research, we analyzed current state-of-theart cyberattack past 10 years major emphasis recent works comparative study identify knowledge gap where work is still needed be done regard category

Language: Английский

---

Enhancing Security and Sensors Emerging Internet of Things (IoT) Technology of Homophone-Based Encryption using MANET‐IoT Networks Technique

Deleted Journal, Journal Year: 2024, Volume and Issue: 20(6s), P. 1345 - 1351

Published: April 29, 2024

The emerging Internet of Things (IoT) technology brings several benefits to the real-time applications smart city project. IoT-enabled technologies are driven by networks like Mobile Ad hoc Network (MANET), Vehicular Hoc (VANET), Wireless Sensor (WSN), Underwater (UWSN), etc. core functionality mainly depends on routing functionality. focus this work is MANET, i.e., IoT-MANET methods. There types research conducted so far efficiency MANET in different areas such as protocol, MAC queue, Routing protocols major domain which performance has been analyzed. have introduced for MANETs. use and selection methods done per application needs. An average On-Demand Vector (AODV) protocol had shown good results under any type network. increase internet usage communication led a significant growth volume data being transmitted over internet, including computer data, text, video, audio, sound, images. However, due limited network capacities, efficient compression needed reduce amount while preserving its quality ensuring timely transmission. Data confidentiality crucial aspect security, homophone-based encryption with (HEC) technique offers viable solution regard.

Language: Английский

---

An Investigation into the Performances of the State-of-the-art Machine Learning Approaches for Various Cyber-attack Detection: A Survey

Published: May 30, 2024

Language: Английский

---