שאלה | 4. [General] Which of the following statements is true about the problems associated with correlated states in DQN algorithms for reinforcement learning (multiple answers may apply)?

4. [General] Which of the following statements is true about the problems associated with correlated states in DQN algorithms for reinforcement learning (multiple answers may apply)?

		a) Correlated states have no impact on the performance of DQN algorithms as long as the neural network architecture is deep enough (assume sufficient training samples for convergence).
		b) Correlated states can improve the performance of DQN algorithms by providing more information about the environment.
		c) Correlated states can lead to overestimation of the Q-values and slow convergence of the algorithm.
		d) Correlated states can lead to underestimation of the Q-values and instability of the algorithm.