Reinforcement learning is a particular area of machine learning concerned with decision making and how software agents take actions in a particular environment in order to maximize a particular function known as reward function .

Theory and Concepts

In Reinforcement Learning an agent receive observations within an environment (also referred as state of the environment), takes actions, and in return it receives rewards which tell the agents how much the action he has taken is good or bad. Its objective is to learn to act in such a way that will maximize the expected long term rewards. We will see later that we can have the cumulative reward that is often time discounted by a factor called gamma. This means that rewards earlier in time weigh more than reward later in time.

Organized in collaboration with Machine Learning Milan on 3 October 2019.
Please complete the following form to access the “A dive into reinforcement learning” free download