Trojan Detection Using Random Forest Algorithm

EasyChair Preprint 8926

Abstract

In recent times Trojan is one of the most common malwares for attack in the PCs. trojans is a one type of malware that create different type of vulnerability that provide exploitation and backdoor entry for attacker. Attacker use trojans for gaining the information from victim’s device like banking or downloader etc. Trojans are small and stealthy and after merging with different file extension the detection are more complicated. These trojans are work efficiently after running the trojan Infected file and make backdoor entry for the attacker. This work makes a model for the detecting the trojans using machine learning. In this technique the detection of trojan is more efficient due to using the random forest algorithm that uses the 20 features for the detecting trojans. The dataset which uses in this paper is generated by the NIST (National Institute of Standards and Technology), CIFAR10 and GTSRB. The overall false acceptance rate is minimum to less than 1% for different type of features (Triggers).

Keyphrases: Computer Security, Cyber Security, Image Processing., Trojan detection, machine learning

@booklet{EasyChair:8926,
  author    = {Sohrab Ansari and Mohit Mohit and Arun Kumr},
  title     = {Trojan Detection Using Random Forest Algorithm},
  howpublished = {EasyChair Preprint 8926},
  year      = {EasyChair, 2022}}