The experimental data and resource of Defect Prediction in Android Binary Executables using Deep Neural Network

Our sampels for defect prediction, including APK and source code file of 5 versions for 15 Android projects respectively. You can get it at http://pan.baidu.com/s/1pKR5N0f

Our source code file of feature & label generator, which is in Analyse. And DNN is our source code for Deep Neaural Network.

• If you want to run them, please confirm your TensorFlow enviroment has already been set up.

This folder contains part of the experimental results and there are all excel file.

1. dataset: All numerical information of defect prediction samples;
2. DNNconf: Results of differnt configuration of our DNN;
3. WP: Results of performance of different classifiers with within-project;
4. CP: Results of performance of different classifiers with cross-project;
5. TIME: The time and space consumption of traing and testing procedure of different classifiers.