DEFENSIVE WALLS AGAINST MACHINE LEARNING ORCHESTRATED ATTACKS

ABSTRACT:

Machine learning (ML)-orchestrated attacks have necessitated advanced security strategies to secure organizations sensitive transaction data with the emergence of sophisticated persistent ML cyberthreat in the digital era. This paper examined the evolving cyber-attacks, emphasizing the need for proactive and dynamic defense strategies. The paper highlights how attackers utilize advanced ML technique, to perpetrate attacks. Thus, optimizing defense strategies is imperative. The paper identified and analysed potential threats, and developed a mitigation strategy using a developed Framework for Securing sensitive Business Data. The paper also developed a defensive strategy against orchestrated ML attacks and proposed an algorithm with a password Generation Checker. The paper proposed a novel process state of art architecture for Confidentiality, Integrity, Availability, Authentication, Authorisation with Non-Repudiation. The proposed scheme is designed to save logs after each attempt/login. When the login information do not match profile, it is flagged and subjected for further review; carefully identifying the nature and pattern of the cyberthreat. The scheme was further validated using mathematical theory. Subsequently, a hybrid approach using Mapping and Elliptic Curve Cryptography (ECC) was developed. The effectiveness of the proposed Scheme was evaluated using Mathematical Key Validation. In addition, the paper developed a password generator and checker using python, educating the organizations on the vital role of passwords to mitigate the persistent cyber threats and presents future directions to addressing evolving threats.

Keywords: Machine-Learning, Elliptic-Curve Cryptography, Cyberthreats, & Defensive-Walls.