In the above writeup, Awkward Saw gives a good description of one kind of machine learning, supervised learning. In fact, the field of machine learning can be described most accurately as a combination of three different types of learning. These are:
- Supervised Learning - In supervised learning, the machine trained on input datapoints that are paired with the correct outputs. The goal is to teach the algorithm to produce the correct output when it is given a new input that it was not trained on. Supervised learning includes both regression and classification. Some supervised learning methods are:
- Unsupervised Learning - During the training portion of unsupervised learning, the machine is given input data that is not paired with output values. In this case, the algorithm builds a mathematical representation of the input data, but cannot match data to a classification. Examples of unsupervised learning are:
- Reinforcement Learning - Reinforcement learning is the type of learning most commonly applied to robotics. In this style of learning, the machine takes input datapoints and uses them to produce actions that affect the machine's environment. Associated with each action is a reward or punishment. The machine's goal is to learn to act in a way that maximizes its reward. Some common methods of reinforcement learning are: