Malware Classification using Deep Learning Methods

With an increase in the fnumber of machines to the internet, the attack surface for cybercriminals has increased multifold, leading to increased risk and damage to the users. One such common attack is due to malicious software (malware) which compromises computers/smart devices, steals confidential information, penetrates networks, and cripples critical infrastructures, etc. The entire cost of malware attacks are projected to be the $3 trillion in 2015 and it is anticipated to rise above $6 trillion by the end of 2021. In order to address and confine the cyber attacks, several approaches such as Intrusion Detection Systems (IDSs) and Intrusion Protection Systems(IPSs), firewalls and antivirus software. These existing malware detection tools, which employ static and dynamic analysis of malware signatures and behaviour patterns, have shown to be inefficient at quickly discovering polymorphic security assaults that haven't been observed before. Also with the maching learning algorithms, feature engineering phase becomes a tedious process with various features present in these datasets. This study incorporates deep learning algorithms to avoid the feature engineering phase and hence, enhance the performance and accuracy of the malware classification.