Auto-exploration for online reinforcement learning

Caleb Ju, Guanghui Lan

The exploration-exploitation dilemma in reinforcement learning (RL) is a fundamental challenge to efficient RL algorithms. Existing algorithms for finite state and action discounte…