What is the purpose of the target network in Deep Q-Learning and how often is it updated?

Machine Learning Hard

Machine Learning — Hard

What is the purpose of the target network in Deep Q-Learning and how often is it updated?

Key points

  • The target network prevents chasing moving targets by holding stable Q-value targets.
  • It is updated periodically to maintain consistency in training.
  • Updating the target network prevents oscillations and ensures more stable learning.

Ready to go further?

Related questions