zohdit / SINVAD

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

SINVAD

This is a replication package for the ICSEW'20 paper SINVAD: Search-based Image Space Navigation for DNN Image Classifier Test Input Generation.

Overview

The ipynb files contain code for the experiments done in the SINVAD paper. They are presented in notebook form to facilitate interactive experimentation. The subdirectories contain code to train the models that are used in SINVAD experiments.

Setup

Set up a python3 environment in your preferred way, then

pip install -r requirements.txt

For each research question's experiments, the prerequisite models are trained using the bash scripts in init_scripts/. To run RQx.ipynb, you would first run

cd init_scripts
sh init_rqx.sh

where x is replaced with the desired RQ number.

Alternatively, to train models for all the RQs, one can run the all_init.sh script in the same directory.

Run

Set up a Jupyter server and run the notebooks.

About


Languages

Language:Jupyter Notebook 95.2%Language:Python 4.7%Language:Shell 0.1%