InsightFinder’s Unsupervised Machine Learning technology stands out because it does not require labeled training data, and accurately identifies anomalies by learning from the data over time. This capability saves engineers time and headache, as they do not have to spend hours training different models as they roll out their AIOps platform.
However, there are times when humans want to add their input and adjust parameters when they know something a machine does not. For example, if a team expects a large increase in storage usage because of a new project, they would not want to be alerted about anomalies that are part of an anticipated project. With InsightFinder, they can easily and precisely dampen certain alerts without missing any true anomalies. In another example, one engineer may want to get alerts about any anomalies for certain key metrics but for certain non-critical metrics only large deviations warrant the alerts. It is impossible to achieve such precise alerting with traditional thresholding approaches. InsightFinder uses its powerful behavior learning techniques to empower engineers with precise alerting.
Primarily using unsupervised machine learning to detect anomalies, while also allowing humans to tweak or adjust thresholds, is at the heart of InsightFinder’s Human Centered AI approach. Unsupervised machine learning saves valuable time and reduces the need for manual tasks. However, it does not replace valuable human knowledge. When human input is needed to make an adjustment, it is easy to incorporate into the system. The combination of these two capabilities is essential to an effective AIOps platform.