/MHM

Generating Adversarial Examples for Holding Robustness of Source Code Processing Models

Primary LanguagePython

MHM

Repository for holding source code of paper Generating Adversarial Examples for Holding Robustness of Source Code Processing Models.

Hierarchy

MHM
├── README.md
├── mhm
    ├── LSTMClassifier			# training LSTM model on poj104 classification task.
    ├── mhm-astnn			# adversarial attack on LSTM classifier.
    ├── mhm-lstm			# training ASTNN model and adversarial attack on it.
    └── poj104				# filelist of poj104 dataset.

Environment

python 3.5 with package requirements:

jupyterlab
numpy
pytorch
pandas
pycparser
gensim
tensorflow

Acknowledgement

Transformer Model is built based on The Annotated Transformer.

ASTNN Model is built based on ASTNN.