This repository contains the implementation of the watermarking algorithm for decision tree ensembles (Random Forests) proposed by Calzavara et. al. in their research paper titled Watermarking Decision Tree Ensembles accepted at the 28th International Conference on Extending Database Technology. This repository also contains the code and the scripts to reproduce the experiments described in the paper.
The artifact is organized in the following folders:
- the datasets folder contains the data used to train the models and the models used in the experiments in the joblib format.
- the src folder contains the python notebooks containing the scripts to run the watermarking algorithm on the three datasets presented in the paper. Moreover, the notebooks contain also the code to reproduce the results presented in the paper;
- the results folder contains a zip folder containing the figures shown in the paper and the intermediate data produced to generate the figures.
Download the repo using git clone <repo_link>.
Here we report some details about the software. Our system used:
- python (3.10)
- some python modules: scikit-learn (1.0.2), numpy (1.26.4), pandas (2.2.1), matplotlib (3.5.1).
- z3-solver==4.13.0.0
At the moment, the implementation of the watermarking algorithm and the other code to reproduce the experiments are inside python notebooks. Precise instructions on how to use the algorithm and reproduce the experiments using independent python scripts will be available after refactoring the code in separate python scripts. Available soon...
The Bibtex entry to cite the paper will be available after the publication of the camera-ready version of the paper.
Thanks to Donald Gera for contributing to the project and preparing the first version of the notebooks.
If you want to ask questions about the artifact and notify bugs, feel free to contact us by sending an email to [email protected].