We are looking at a particular subclass of challenges where we want to learn an input-to-target mapping. This subclass is generally referred to as supervised machine learning. The word supervised denotes that we have target for each input. Unsupervised machine learning includes challenges such as trying to cluster text, where we do not have a target.
To do any supervised machine learning, we need the following in place:
- Input Data: Anything ranging from past stock performance to your vacation pictures
- Target: Examples of the expected output
- A way to measure whether the algorithm is doing a good job: This is necessary to determine the distance between the algorithm's current output and its expected output
The preceding components ...
