Machine learning based malwares detection

Nowadays, one of the most important problems in terms of information security is to detect malware. Hackers often aim to exploit systems using malicious software. For this reason, the detection and prevention of malware is critical to information security. In general, the behavior of a malware is determined and detected using a signature-based system. However, theoretically malicious software can have endless behavior. It is known that detection of malware by using deterministic systems is difficult and will be more difficult. Artificial intelligence and machine learning methods are used to solve this problem. Recently, deep learning has been seen as a phenomenon of machine learning and artificial intelligence and its use is becoming widespread. Deep learning systems are used in information security as well as in areas such as image, voice and text recognition. In particular, deep learning methods are often used in malware detection, but deep learning methods remain at 85% recognition rates in heterogeneous malware data sets. In this thesis, the tools that hackers frequently, the effects of malware and anti-malware solutions have been investigated and their effects on malware recognition have been investigated and thus, it has been aimed to recognize such malware with higher accuracy by using new generation feature extraction and learning methods.