When and Why Test Generators for Deep Learning Produce Invalid Inputs: an Empirical Study

Open in new window