Combining policy gradient and Q-learning
This addresses a bottleneck in reinforcement learning for more efficient and stable training, though it is incremental as it builds on existing methods.
The paper tackles the problem of vanilla policy gradient methods being on-policy and inefficient with off-policy data by introducing PGQL, a technique that combines policy gradient with off-policy Q-learning using a replay buffer. It demonstrates improved data efficiency and stability, achieving performance exceeding A3C and Q-learning on the full suite of Atari games.
Policy gradient is an efficient technique for improving a policy in a reinforcement learning setting. However, vanilla online variants are on-policy only and not able to take advantage of off-policy data. In this paper we describe a new technique that combines policy gradient with off-policy Q-learning, drawing experience from a replay buffer. This is motivated by making a connection between the fixed points of the regularized policy gradient algorithm and the Q-values. This connection allows us to estimate the Q-values from the action preferences of the policy, to which we apply Q-learning updates. We refer to the new technique as 'PGQL', for policy gradient and Q-learning. We also establish an equivalency between action-value fitting techniques and actor-critic algorithms, showing that regularized policy gradient techniques can be interpreted as advantage function learning algorithms. We conclude with some numerical examples that demonstrate improved data efficiency and stability of PGQL. In particular, we tested PGQL on the full suite of Atari games and achieved performance exceeding that of both asynchronous advantage actor-critic (A3C) and Q-learning.