Given a series of inputs x1, x2, ... to a system:

  • Supervised learning: The system is also given the desired outputs y1, y2, . . . The aim is to learn to produce the correct output given a new input.
  • Unsupervised learning: The aim is to build representations of inputs so that it can produce the desired outputs without further information.
  • Reinforcement learning: The system's outputs change the states of the world, leading to rewards (or punishments). The aim is to learn to act in a way that maximises rewards in the long term.