As the Android operating system continues to grow in popularity, malware attacks on the platform evolve in complexity accordingly. In this project, I developed a Convolutional Neural Network (CNN) model to detect whether some code is either malicious or benign.
For this project I used the Drebin-215 dataset provided and used by the DroidFusion researchers ( https://www.sec.cs.tu- bs.de/~danarp/drebin/index.html ). The Drebin-215 dataset has a total of 15036 samples, with 5560 recorded as malware and the remaining as benign. It has 215 features and 2 classes, malware and benign.
After completing my model using Keras, I was able to achieve an accuracy score of 97.8%.