iNNE

iNNE - Isolation-based anomaly detection using nearest-neighbor ensembles.

Based on the paper:

Tharindu R., et al. Isolation-based anomaly detection using nearest-neighbor ensembles. Computational Intelligence (2018)

Matlab code of iNNE:

https://github.com/zhuye88/iNNE

Introduction to the paper:

https://www.jianshu.com/p/379a5898beb6

Abstract of the paper:

The first successful isolation-based anomaly detector, ie, iForest, uses trees as a means to perform isolation. Although it has been shown to have advantages over existing anomaly detectors, we have identified 4 weaknesses, ie, its inability to detect local anomalies, anomalies with a high percentage of irrelevant attributes, anomalies that are masked by axis-parallel clusters, and anomalies in multimodal data sets. To overcome these weaknesses, this paper shows that an alternative isolation mechanism is required and thus presents iNNE or isolation using Nearest Neighbor Ensemble. Although relying on nearest neighbors, iNNE runs significantly faster than the existing nearest neighbor-based methods such as the local outlier factor, especially in data sets having thousands of dimensions or millions of instances. This is because the proposed method has linear time complexity and constant space complexity.

Documentation, including tutorials, are available on ReadTheDocs at https://inne.readthedocs.io.

Installing

PyPI install, presuming you have an up to date pip.

pip install inne

For a manual install of the latest code directly from GitHub:

pip install git+https://github.com/xhan97/inne.git

Alternatively download the package, install requirements, and manually run the installer:

wget https://codeload.github.com/xhan97/inne/zip/refs/heads/master
unzip inne-master.zip
rm inne-master.zip
cd inne-master

pip install -r requirements.txt

python setup.py install

How to use iNNE

The inne package inherits from sklearn classes, and thus drops in neatly next to other sklearn with an identical calling API. Similarly it supports input in a variety of formats: an array (or pandas dataframe) of shape (num_samples x num_features).

from inne import IsolationNNE
from sklearn.datasets import make_blobs

data, _ = make_blobs(1000)

clf = IsolationNNE(n_estimators=200, max_samples=16)
clf.fit(data)
anomaly_labels = clf.predict(data)

Running the Tests

The package tests can be run after installation using the command:

pip install pytest

or, if pytest is installed:

pytest  inne/tests

If one or more of the tests fail, please report a bug at https://github.com/xhan97/inne/issues

Python Version

Python 3 is recommend the better option if it is available to you.

Citing

If you have used this codebase in a scientific publication and wish to cite it, please use the following publication (Bibtex format):

@article{bandaragoda2018isolation,
        title={Isolation-based anomaly detection using nearest-neighbor ensembles},
        author={Bandaragoda, Tharindu R and Ting, Kai Ming and Albrecht, David and Liu, Fei Tony and Zhu, Ye and Wells, Jonathan R},
        journal={Computational Intelligence},
        volume={34},
        number={4},
        pages={968-998},
        year={2018},
        publisher={Wiley Online Library} }

How to contribute

Thanks for your interest in contributing to inne. A guide is shown in inne's Documentation.

License

BSD license