A Deep Neural Network Model for Malware Detection
Fatima Bourebaa, Mohamed Benmohammed · DergiPark (Istanbul University) · 2021
Parallel to the adoption of mobile technology in our daily lives, there is a growing and increasing proliferation of cyber frauds and malicious content.Mobile malware can exploit the vulnerabilities of the device, modify, disclose or erase confidential data, such as credit card numbers, passwords, medical data, contacts, or even block the device asking for a ransom.In this paper, we leverage the possibilities of deep fully-connected neural networks, using permissions and Application Programming Interfaces APIs as features, to automatically and efficiently detect Android malware.We achieved a score of 88.9% using a feedforward of 128x128x1, 2-hidden layers configuration.