ADVANCED MALICIOUS SOFTWARE DETECTION USING DNN

Abstract

The special component of malicious software analysis is advanced malicious software analysis which implicates interested the main framework of malicious software that can be executed after executing it and aggressive malicious software investigation depend on inquisitive of the practice of malicious software after running it in a composed habitat. Advanced malicious software analysis is usually performed by contemporary anti-malicious software operating systems using signature-based analysis. The purpose of this research is to propose also decide a DNN for the progressive identification of portable files to study the features of portable executable malicious software to minimize the occurrence of distorted likeness when aware of advanced malicious software. The model proposed in this study is a NN with a Dropout model contrary to a resolution tree model to examine how well it performs in detecting real malicious PE files. Setup-skeptic methods are used to extract features from files. The dataset is used to train the proposed approach and measure outcomes by alternative common malicious software datasets. The results from this study illustrate that the use of simple DNNs to study PE vector elements is not only efficient but more fewer system comprehensive than the traditional interested disclosure approach. The model proposed in this study achieves an A-UC of ninety-nine point eight with ninety accurate specifics at one percent inaccurate specific on the R-OC curve. For shows that this model has the potential to complement or replace conventional anti-malicious software operating systems so for future research, it is proposed to implement this model practically.