What type of study is this?

September 10, 2025Open Access

Combative Machine Learning Approaches for Solving Bot-Driven Cyber Threats in Social Networking Platforms

SNS. N.Andhra University SKSenthil Kumar KSRM Institute of Science and Technology

Key Points

The proposed framework significantly improves detection accuracy compared to traditional machine learning models.
Adversarial machine learning techniques were applied to enhance the robustness against evasion and poisoning attacks.
This research highlights the need for advanced detection systems in light of escalating bot-driven cyber threats.
Benchmark datasets showed that traditional models are ineffective against sophisticated bot-driven threats.

Abstract

Social networking platforms have become prime targets for bot-driven cyber threats. The unpredictable expansion of social networking platforms has resulted in bot-driven cyber threats, which includes spreading misinformation, spam, phishing, and automated account takeovers. Traditional machine learning models are not very effective as they are susceptible to combative attacks. This paper proposes a novel robust detection framework using adversarial machine learning (AML) techniques to enhance the robustness of bot detection systems. Attacks involving evasion and Poisoning, directed at ML-based security systems, have been observed, and some defensive strategies help counter these types of attacks. Experimental results on benchmark datasets demonstrate significant improvements in detection accuracy compared to traditional models.

AI से पूछें

Bookmark

View Full Paper