Deep Learning is simply a way of deepening the process by which a machine learning application support increased accuracy of a prediction, while reducing the error drastically. This learning process can be applied on unstructured and structured datasets.
Deep Learning using Reinforcement Learning is more of using live unstructured data to ensure back-propagation and repetitive analysis of an Agent within an Environment to establish a learning and a way of working for the Agent.
Introduction to Cartpole: This virtual project allows the use of reinforcement learning to keep a pole upright and accurately above the cart without falling.