Designing a new model for Trojan horse detection using sequential minimal optimization

• Main Authors: M.M., Saudi, A.M., Abuzaid, B.M., Taib, Z.H., Abdullah
• Format: Conference Paper
• Language: en_US
• Published: Springer Verlag 2015
• Subjects: Artificial intelligence; Learning algorithms; Learning systems; Optimization; Error detection; Automated analysis; False positive rates; Sequential minimal optimization; Sequential minimal optimization algorithms; Trojan detections; Trojan Horse attacks; Trojan horse detection; True positive rates; Malwares; Trojan horse; Malware
• Online Access: http://ddms.usim.edu.my/handle/123456789/9177
• Institution: Universiti Sains Islam Malaysia

Malwares attack such as by the worm, virus, trojan horse and botnet have caused lots of troublesome for many organisations and users which lead to the cybercrime. Living in a cyber world, being infected by these malwares becoming more common. Nowadays the malwares attack especially by the trojan horse is becoming more sophisticated and intelligent, makes it is harder to be detected than before. Therefore, in this research paper, a new model called Efficient Trojan Detection Model (ETDMo) is built to detect trojan horse attacks more efficiently. In this model, the static, dynamic and automated analyses were conducted and the machine learning algorithms were applied to optimize the performance. Based on the experiment conducted, the Sequential Minimal Optimization (SMO) algorithm has outperformed other machine learning algorithms with 98.2 % of true positive rate and with 1.7 % of false positive rate.