Anomaly Detection in Data streams using MOA
Year: 2021
Type: article
Abstract: Anomaly means anything which deviates from normal. It can be a credit card fraud or sensor alarm or a signal from a condition monitoring device. A problem like anomaly arises when we try to monitor the unusual behaviour of a machine. More number of outliers means the machine needs to be inspected. Anomaly detection in static data can be entirely different from that of streaming data.
We have some issues in anomaly detection in streaming data when compared to static data. If any off – line algorithms attempt to find anomalies in streams, it has to store the entire stream for analysis. So, there is a high probability that it will run out of memory space.
Also, streams can be infinite and evolving over a period of time because of which maintenance of high detection accuracy becomes almost impossible. In this paper we will discuss about anomaly detection in data streams and using MOA (Massive Online Analysis) tool we will analyse which algorithm derives best results.
Author: G. Sandhya Madhuri
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W3190734578