What aspect of data does novelty detection focus on?

Novelty detection focuses on identifying previously unknown patterns within a dataset. This technique involves analyzing the data to discover observations that significantly differ from the established patterns or typical behavior of the majority of data points. It is particularly useful in scenarios where new and unseen data might not conform to the existing model or distribution, such as fraud detection, network intrusion detection, and identifying anomalies in various domains.

By concentrating on previously unknown patterns, this approach allows for the discovery of outliers or unusual occurrences that were not classified during the training phase, which is essential in many practical applications where such anomalies can have significant implications. The emphasis is on exploring the data for these novel insights rather than simply relying on pre-existing knowledge or categorization of typical patterns. This distinct focus is why identifying previously unknown patterns accurately encapsulates the essence of novelty detection.