A decision process: A set of calculations or steps that takes in data and makes a guess about the pattern your algorithm is trying to find.

An error function: A way to measure how good a guess was by comparing it to known examples, if they are available. Did the decision process get it right? If it didn't, how can you measure “how bad” the mistake was?

An updating or optimization process: A process where the algorithm examines the miss and then updates how the decision process arrives at the final decision so that next time, the miss won't be as large.

This type of machine learning is like learning with a teacher. The computer is given data that includes both the input (like questions) and the correct answers (called labels). The goal is to teach the computer to make predictions or decisions based on examples.

Example:

Imagine a folder of images where each image is labeled with a category (e.g., “cat,” “dog”). The computer analyzes these labeled examples and learns to recognize animals based on features like fur patterns or ear shapes. Later, when shown an unlabeled image, it predicts whether it’s a cat or a dog.

This type of machine learning has no teacher. The computer is only given data without any answers or labels. It tries to find patterns, groupings, or structures on its own.

Example:

Imagine a folder of images with no labels provided. The computer analyzes the images and groups them based on similarities, such as fur patterns or ear shapes. It might create clusters of images that look like cats and others that look like dogs, without knowing the actual labels.