Thesis Project with the titel: "Interpretation of Neural Network Classifiers using Rule-based Models"
The source code consists of several scripts used to achieve named goal.
The main idea is to create a Rough Set model that explains data which was labelled by a Neural Network.
In addition, an analysis of the rules containing wrongly classified objects in their support set allows to gain insights in why the Neural Network makes mistakes or where the feature space must be enlarged.
Data and Results are ordered based on the use cases presented in the thesis. For the source code an additional ReadMe is contained in the directory.