What is the problem of finding hidden structure in unlabeled data called?

The correct term to describe the problem of finding a hidden structure in unlabeled data is called unsupervised learning. It is a machine learning method that allows users not to master too much information on the model they are working on.

When you compare supervised learning with unsupervised learning, you will realize that more tasks can be accomplished with the latter compared to the former. This is usually because you are dealing majorly with unlabeled data. Algorithms such as anomaly detection and clustering are usually encountered under unsupervised learning.

A perfect example to explain what unsupervised learning means is a scenario of a kid whose toy car was stolen but still able to operate another one given to him by his parents to a particular extent. If anybody had told him how to operate it, it would, however, be supervised learning.