A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
Unknown
Conservative Q-Learning (CQL) is an algorithmic framework for offline reinforcement learning that learns a Q-function to address the challenge of distribution shift.
Ilya Kostrikov, Ashvin Nair, S. Levine
Introduces Implicit Q-Learning (IQL), an offline RL algorithm that avoids evaluating actions outside the dataset by using expectile regression and does not require unseen actions.
Rashid Ali, Nurullah Shahin, Yousaf Bin Zikria, et al.
Proposes a deep reinforcement learning (Q-learning) based resource allocation mechanism (iQRA) to optimize channel observation-based MAC protocols in dense WLANs.
Hado van Hasselt, Arthur Guez, David Silver
This paper shows that DQN overestimates action values in Atari games and proposes a Double DQN algorithm that reduces overestimation and improves performance.