Intuition
Deep Convolutional Q-Learning is a reinforcement learning method that visually perceives an image to understand what to do next to maximize the reward to achieve a certain task.
Let’s say we are playing Mario. We can use DCQLearning to have the model learn which way to go to get to the finish line.
The model first receives the display image of the game, applies convolutions and pooling, and connects the flattened output to a neural network. The network will then output an action so that it can get high rewards when considering getting to the finish line. Finally, the model will modify the next step based on that reward.
