Double DQN
An improved variant of DQN that decouples action selection from value evaluation to mitigate overestimation of Q-values and improve training stability.
Timeline (last 90 days)
Double DQN is an improved version of DQN proposed by Hasselt et al. in 2016, which mitigates the Q-value overestimation problem by using the online network to select actions and the target network to evaluate Q-values.
Double DQN decouples action selection from value estimation by using the online network to select actions and the target network to evaluate Q-values, thereby alleviating the problem of Q-value overestimation.
Double DQN 用于解决过估计，Dueling DQN 分离状态价值和优势函数，Prioritized Experience Replay 优先采样TD误差大的经验
All Facts (3)
Double DQN is an improved version of DQN proposed by Hasselt et al. in 2016, which mitigates the Q-value overestimation problem by using the online network to select actions and the target network to evaluate Q-values.
50%UnverifiedDouble DQN decouples action selection from value estimation by using the online network to select actions and the target network to evaluate Q-values, thereby alleviating the problem of Q-value overestimation.
50%UnverifiedDouble DQN 用于解决过估计，Dueling DQN 分离状态价值和优势函数，Prioritized Experience Replay 优先采样TD误差大的经验
50%