If you have a baby, you know already machine learning

Machine learning is somewhat similar to the way humans learn. An individual may learn under the supervision of his parents, teachers, or anyone who has knowledge beyond his own. For example, when a child points to a "ball" and asks his mother about it, and she answers that it is a ball, the child then becomes able to recognize balls of different shapes and this is what is called in machine learning supervised learning.

You may have noticed that when the child plays with different types of toys, he may try to divide them into groups according to the similarity that he sees, which may be the color, the geometric shape or both, and this without prior knowledge of those toys. We call this type of learning unsupervised learning.

The last type of machine learning is reinforcement learning, and it is the dominant type in the learning process, as you learn through experience, that is, you perform an action and according to its result, i.e. the size of the reward you get, you decide whether what you have done is worth repeating or not. These rewards are predetermined. In large areas and without prior knowledge, it is difficult to inventory all possible actions and thus determine the rewards, so we resort to a special type of reinforcement learning, which is curiosity-based learning, where a person's curiosity is what drives him to learn and try everything he sees new. This type of learning is observed in infants, as the latter tries to identify his surroundings and himself as well by moving his hands and looking at the results of this movement, and he might have succeeded in clapping.

Humans learn in this way without any interference, but machines need algorithms to do the job.