Deep Q-learning to play Space Invaders game

In the previous section, we used Deep Q-learning to play the Cart-Pole game. In this section, we will leverage Deep Q-learning to play Space Invaders, which is a more complex environment than Cart-Pole.

A sample screenshot of the Space Invaders game looks as follows:

source: https://gym.openai.com/envs/SpaceInvaders-v0/

The objective of this exercise is to maximize the score obtained in a single game.

Getting ready

The strategy that we'll adopt to build an agent that is able to maximize the score is as follows:

• Initialize the environment of the Space Invaders-Atari2600 game.
• Preprocess the image frame:
  • Remove pixels that do not necessarily impact the action prediction
    • For...