Effectively spotting anomalies in network or application operations can be challenging in very large, complex networks, making it difficult to be alerted to their presence in order to take action to remediate such anomalies. Proper anomaly detection is impeded by too much data from too many disparate sources, which may manifest in too many different ways on various network devices ("the curse of dimensionality" familiar to many machine learning (ML) practitioners).

Proposed herein is a novel implementation of a Long Short-Term Memory based Variational Autoencoder (LSTM-VAE) to detect such anomalies, and an associated visualization technique to display them to the network manager for subsequent remediation. By so doing, the described technique provide a novel method of anomaly detection in large, complex networks in a way not otherwise possible.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.