Implement Rainbow DQN for solving Markov Decision Processes

THIS CODE IS FOR RESEARCH PURPOSES ONLY AND IS NOT TO BE USED FOR TRADING OR FINANCIAL ADVICE!

• Solve the example control task "Qbert"

• Solve the example control task "Trading Bot"

• Implements Rainbow Deep Q-Network algorithm:

  • Double DQN
  • Dueling DQN
  • Prioritized Experience Replay
  • Noisy DQN
  • N-Step DQN
  • Distributional DQN

References

• Rainbow DQN whitepaper
• Advanced Deep Q Networks course
• Trading Bot example