Anomaly Detection - Unsupervised vs Supervised Learning - where does it belong?

Course Q&A Machine Learning Specialization Supervised ML: Regression and Classification
, ,
1

Dear ML enthusiasts,

Why is Anomaly Detection deliberately (or not?) presented as Unsupervised Learning task? (from the beginning of ML Specialization course…)

Suggestion made like that might be misleading because Anomaly Detection task(s) can be also implemented with the aid of Supervised Learning, right?

Could be training materials supplemented with the explanation on why “Anomaly Detection” falls into “Supervised Learning” category? Otherwise, if “not”, could be “heads-up” provided to clear the misunderstanding?

What’s your dear opinion MLEs?

Kind regards,
an MLE :slight_smile:

2

Anomaly detection relies only on statistical analysis of a labeled sample of the data.
There are no weights or training - only adjusting an anomaly threshold.

So it’s not really a classical “supervised learning” method.

2 Likes