The goal of this thesis was to evaluate existing machine learning solutions upon a data set of cryptographic misuses. The result should enable machine learning algorithms to label cryptographic usages as secure or insecure.

While the data set was created with it's applicability for machine learning, the evaluation was missing. Thus, the aim of the thesis is to use the data set to train a machine learning model to learn crypto usages rules without expert input.

This thesis discussed the influence of the data set on the results and compared the machine learning algorithms Naive Bayes and SVM.

Publications

• Parisa Rashidirad: Machine Learning on Code Using a Dataset of Parametric Crypto Misuses.