The folder exp-DLF-FSE contains the data resulting from the experimental procedure described in the paper.
The metrics can be computed by running the python script evaluation_radius_DLF.py provided in this folder.
The output of these scripts is a csv file containing the data reported in the paper. The expected output is reported in Table3-DLF.csv.
DLFuzz is the tool proposed in the paper DLFuzz: differential fuzzing testing of deep learning systems by J. Guo et al. The code can be obtained from https://github.com/turned2670/DLFuzz.