Online evolutionary Spiking Neural Network unsupervised anomaly detector

Description

Python implementation of OeSNN-UAD model. Model finds anomalies in one dimensional data streams. Theoretical basics about this model could be find here: https://arxiv.org/pdf/1912.08785.pdf.

Dependencies

• numpy
• PyEMD (pip install EMD-signal)

Package instalation

To install package for python you should type in terminal:

    pip install OeSNN-AD

Usage

Our model require from data stream to be numpy array. Additional model parameters are passed as arguments in object constructor.

The following code snippet shows package basic usage.

    from oesnn_ad import OeSNNAD
    import numpy as np

    data_stream = np.array([1, 2, 3, 4, 5])
    model = oesnn_ad(data_stream)

    results = model.predict()

Parameters

The following table shows model parameters and their values range.