Reinforcement learning algorithms use a goal-oriented approach to learn with a focus on maximising a complex goal or cumulative reward.