Automatic Model-agnostic Regional Rule Extraction (AMORE) is the implementation of our paper "Enabling Regional Explainability by Automatic and Model-agnostic Rule Extraction".
These instructions will help you install a virtual environment to reproduce our experimental results in the paper.
- Python 3.9.15
- virtualenv/ pyenv / conda (Choose according to the environment setup you prefer)
-
Clone the repository
git clone https://github.com/yc14600/AMORE.git cd AMORE -
Using
requirements.txt(for virtualenv or pyenv) to create virtual environment1). Create a virtual environment
Create a virtual environment by virtualenv or pyenv with specified python version 3.9.15, for example:
pyenv virtualenv 3.9.15 amore_venv
2). Activate the virtual environment
pyenv activate amore_venv
3). Install the dependencies
pip install -r requirements.txt
-
Using
environment.yml(Conda)1). Create a Conda environment
conda env create -f amore_venv.yml
2). Activate the Conda environment
conda activate amore_venv
All experiments are Ipython notebooks in the "experiments" folder. You can run the notebooks in the following way:
After activating the virtual environment, first register the virtual environment to ipython kernel:
python -m ipykernel install --user --name=amore_venvThen run jupyter notebook in the terminal and open the notebooks in the browser. For example, run the following commands in the terminal:
cd experiments
jupyter notebookTo specify the virtual environment for running a notebook file, click the Kernel tab and then click change kernel in the list, choose the kernel amore_venv.
There are two datasets from Kaggle website, which need to be downloaded first before running the corresponding notebook. The links are as follows:
Other datasets can be downloaded automatically by running the notebook first time.
This project is licensed under the CC-BY-4.0 License - see the LICENSE.md file for details