Machine Learning

Unsupervised Learning

Unsupervised learning is about finding structure hidden in collections of unlabeled data.

One of the things is that in Supervised Learning, we have a Loss Function, which uses some sort of distance metric that we want to try to minimize. Usually, it is very straightforward and we just use either L1 or L2 distance.

However, in unsupervised, it seems that this is discussed more. They talk about the Curse of Dimensionality.

I thought this is a problem in Supervised Learning too? No, because you labels for them, so it is very easy to compare. But for Supervised Learning too? No, because you labels for them, so it is very easy to compare. But for Unsupervised Learning, you don’t necessarily know.

Self-Supervised Learning

Concepts