UCB EXPLORATION VIA Q-ENSEMBLES
arXiv: Learning, 2018.
We show how an ensemble of $Q^*$-functions can be leveraged for more effective exploration in deep reinforcement learning. We build on well established algorithms from the bandit setting, and adapt them to the $Q$-learning setting. We propose an exploration strategy based on upper-confidence bounds (UCB). Our experiments show significant ...More
PPT (Upload PPT)