Looks to be a useful thing, depending on your goals and definitions of anomaly. Like the definition of usage cases.
Using AI to Find Anomalies Hiding in Massive Datasets
MIT News, Adam Zewe, February 25, 2022
A machine learning model developed by researchers at the MIT-IBM Watson AI Lab can identify anomalies in massive data streams automatically, in real time. The researchers used artificial intelligence (AI) to model the interconnectedness of the power grid to better detect glitches, without needing annotated data on power grid anomalies for training. This means the model could be applied in real-world situations where the availability of high-quality, labeled datasets is limited. The researchers indicated the model could be applied to other circumstances where a large number of interconnected sensors collect and report data. The framework was tested on power grid, traffic, and water system datasets that contained anomalies previously identified by humans, and outperformed the baselines by identifying a higher percentage of true anomalies in each dataset.
No comments:
Post a Comment