Abstract
Mutation testing can be used to assess the fault-detection capabilities of agiven test suite. To this aim, two characteristics of mutation testingframeworks are of paramount importance: (i) they should generate mutants thatare representative of real faults; and (ii) they should provide a complete toolchain able to automatically generate, inject, and test the mutants. To addressthe first point, we recently proposed an approach using a Recurrent NeuralNetwork Encoder-Decoder architecture to learn mutants from ~787k faults minedfrom real programs. The empirical evaluation of this approach confirmed itsability to generate mutants representative of real faults. In this paper, weaddress the second point, presenting DeepMutation, a tool wrapping our deeplearning model into a fully automated tool chain able to generate, inject, andtest mutants learned from real faults. Video:https://sites.google.com/view/learning-mutation/deepmutation