DeepMutaion++ is a mutation testing framework for deep learning systems implemented in Python3.6 with TensorFlow1.13 and Keras2.2.4 . DeepMutaion++ supports both FNN and RNN mutants generation and can be used to analyze analyze the robustness of a DNN model against the input as a whole. Moreover, it allows to identify the vulnerable segment of a sequential input.
├── cnn_mutation # cnn mutation folder
│ ├── src # code folder
│ │ ├── cnn_operator.py # define cnn mutation operators
│ │ ├── data_sort.py # sort data by KS1
│ │ ├── generator.py # mutants generator
│ │ ├── run.sh # shell for test
│ │ ├── utils.py # some useful functions
│ │ ├── __init__.py #
│ ├── __init__.py #
│ ├── README.md #
├── data # save some test data
├── img # save imgs
├── models # save some test models
├── result # save some test results
├── rnn_mutation # rnn mutation folder
│ ├── src # code folder
│ │ ├── keras # keras library
│ │ │ ├── ... #
│ │ ├── __init__.py #
│ │ ├── gru_operator.py # define gru mutation operators
│ │ ├── lstm_operator.py # define lstm mutation operators
│ │ ├── run.sh # shell for test
│ │ ├── runner.py # mutants generator
│ │ ├── sort_segment.py # sort segment by KS2
│ │ ├── state_save.py # save state
│ │ ├── utils.py # some useful functions
│ ├── README.md #
│ ├── __init__.py #
├── LICENSE # MIT license
├── README.md #
├── test.py # for my test
├── utils.py # some useful functions
