Hybrid Model Approach for Real-Time Acoustic Anomaly Detection using Time Series

Detecting anomalous behaviours help to find new knowledge of a given phenomenon since anomaly behaviours rarely occurred and give better insights to prevent those instances in the future. Usually, anomaly behaviours happen unexpectedly due to the unconscious nature of the objects. This project will directly or as a first insight helps to prevent happening severe anomaly behaviours generated in real-time via helping to diagnose them as early as possible.

Thus far, an increasing number of machine learning and deep learning algorithms have been developed to detect anomalies, but only a few applications focus on detecting acoustic anomalies
in real time. With respect to that, the core contribution of this research project is given by delivering an improved hybrid model architecture to capture more time series features for Real-Time Acoustic Anomaly Detection and enable it to use in the above
scenarios effectively.