对抗性MDP策略优化弥合horizon gap研究

1. 12:00
   arXiv · Statistics Machine Learning

   对抗性 MDP 策略优化中 horizon gap 的弥合

   研究针对带对抗损失与 bandit feedback 的在线 episodic tabular MDP 策略优化，用 regularized Q-functions 联合控制所有 state-action pair 的局部更新稳定性，弥合了与 occupancy-measure 算法相差 horizon H 的 regret 差距。

Current heat 9·Comparable peak 10（Oct 9）·Comparable change over 24 hours –

The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.