In the lectures, both in the planning and learning settings, we only considered MDPS in which the set of actions is discrete and enumerable. In practice one often finds tasks that have a continuous action space (with an infinite number of actions). Take Sarsa as an illustrative method. What major changes would be needed in practice to apply the method on a task with a continuous action space? Your answer should take no more than 5-6 lines. [2 marks]