UCB Exploration via Q-Ensembles

2017-06-05ICLR 2018Unverified0· sign in to hype

Richard Y. Chen, Szymon Sidor, Pieter Abbeel, John Schulman

Unverified — Be the first to reproduce this paper.

Abstract

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 gains on the Atari benchmark.

Tasks

Deep Reinforcement Learning Q-Learning reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)

UCB Exploration via Q-Ensembles

Abstract

Tasks

Reproductions