Reinforcement learning is a branch of machine learning that deals with how to take an agent's actions and figure out how those actions have affected the reward it receives from performing that action.

There are many different kinds of reinforcement learning algorithms, including value iteration, policy gradient, and Monte Carlo methods. The most popular of these is value iteration, which uses an optimization algorithm to determine the best action a robot should take given its current state, current environment, and any other information it has access to (such as what kind of outcome it wants).

And also deep reinforcement learning is a technique that uses deep neural networks to learn from experience. When you're using deep reinforcement learning, you have the ability to train your machine learning models with the help of a human trainer who provides feedback on how well the model is performing.

The goal of this technique is to develop an algorithm that can learn how to perform a task without being explicitly programmed. Deep reinforcement learning works by creating an artificial neural network that learns to achieve a result through trial and error, over time. This type of algorithm has a large number of layers and nodes that allow it to learn from experience as well as from other similar systems or situations.