On the Role of Computation in Reinforcement Learning

How does the amount of compute available to a reinforcement learning (RL) policy affect its learning? Can policies using a fixed amount of parameters, still benefit from additional compute? The standard RL framework does not provide a language to answer these questions formally. Empirically, deep RL policies are often parameterized as neural networks with static architectures, conflating the amount of compute and the number of parameters. In this paper, we formalize compute bounded policies and prove that policies which use more compute can solve problems and generalize to longer-horizon tasks that are outside the scope of policies with less compute. Building on prior work in algorithmic learning and model-free planning, we propose a minimal architecture that can use a variable amount of compute. Our experiments complement our theory. On a set 31 different tasks spanning online and offline RL, we show that $(1)$ this architecture achieves stronger performance simply by using more compute, and $(2)$ stronger generalization on longer-horizon test tasks compared to standard feedforward networks or deep residual network using upto 5 times more parameters.

Reinforcement Learning (RL) policies perform some amount of computation to output an action after receiving an observation. Prior work in RL typically conflate the amount of compute that an RL policy uses with the number of parameters in that policy. In this paper, we take a principled approach and investigate the role of the computational ability of a policy towards its performance. We prove two primary theorems that show the benefit of using policies that can use more compute. We also propose a simple compute adaptive policy architecture that outperforms much larger policy architectures on some combinatorial decision making problems.