Optimum-Path Forest Stacking-based Ensemble for Intrusion Detection

This repository holds all the necessary code to run the very-same experiments described in the paper "Optimum-Path Forest Stacking-based Ensemble for Intrusion Detection".

References

If you use our work to fulfill any of your needs, please cite us:

@article{Bertoni:21,
  author = {Mateus A. Bertoni and Gustavo H. de Rosa and Jose R. F. Brega},
  title = {Optimum-path forest stacking-based ensemble for intrusion detection},
  journal = {Evolutionary Intelligence},
  doi = {10.1007/s12065-021-00609-7},
  year = {2021},
  month = may,
  publisher = {Springer Science and Business Media {LLC}}
}

Structure

• data/
  • nsl-kdd.tar.gz: A compressed file containing the NSL-KDD dataset;
  • unespy.tar.gz: A compressed file containing the uneSPY dataset;
• models/
  • stacking.py: A class that implements a stacking-based ensemble;
• utils/
  • classifiers.py: Wraps the classifiers classes;
  • load.py: Loads the dataset according to desired format;
  • wrapper.py: Wraps the classification tasks.

How-to-Use

There are 4 simple steps in order to accomplish the same experiments described in the paper:

• Install the requirements;
• Perform the single classifier classification (using k-folds or split);
• Perform the stacking-based ensemble classification;
• Process post-classification information for further comparison;

Installation

Please install all the pre-needed requirements using:

pip install -r requirements.txt

Classification with k-folds or split

Our first classification script is to use the k-folds cross validation to mitigate the randomness of the dataset. With that in mind, just run the following script with the input arguments:

python single_classifier_kfolds.py -h

Our second classification script is to use the traditional training/testing set split validation to mitigate the randomness of the dataset. With that in mind, just run the following script with the input arguments:

python single_classifier_split.py -h

Note that it will output a helper file in order to assist in choosing the correct arguments for the script.

Classification with stacking-based ensemble

Additionally, we offer a stacking-based ensemble to construct more robust classifiers. With that in mind, just run the following script with the input arguments:

python stacking_classifier.py -h

Note that it will output a helper file in order to assist in choosing the correct arguments for the script.

Post-classification processing

Finally, after concluding the classification step, it is now possible to load back the metrics found during the classification procedure and output them to more-readable files. Run the following script in order to fulfill that purpose:

python process_single_metrics_file.py -h

Also, if one have performed an ensemble-based classification, please use the following script:

python process_ensemble_metrics_file.py -h

Note that the classification process will always output a .pkl file, while the other scripts will output a *.txt file.