Imitation-Projected Programmatic Reinforcement Learning
This addresses the problem of designing rigorous learning approaches for programmatic policies, which are more interpretable and verifiable than neural policies, but is incremental as it builds on existing deep policy gradient and program synthesis methods.
The paper tackles the challenge of learning interpretable and verifiable programmatic policies in reinforcement learning by introducing PROPEL, a meta-algorithm that combines gradient-based optimization with program synthesis via imitation learning, achieving significant performance improvements over state-of-the-art methods in continuous control domains.
We study the problem of programmatic reinforcement learning, in which policies are represented as short programs in a symbolic language. Programmatic policies can be more interpretable, generalizable, and amenable to formal verification than neural policies; however, designing rigorous learning approaches for such policies remains a challenge. Our approach to this challenge -- a meta-algorithm called PROPEL -- is based on three insights. First, we view our learning task as optimization in policy space, modulo the constraint that the desired policy has a programmatic representation, and solve this optimization problem using a form of mirror descent that takes a gradient step into the unconstrained policy space and then projects back onto the constrained space. Second, we view the unconstrained policy space as mixing neural and programmatic representations, which enables employing state-of-the-art deep policy gradient approaches. Third, we cast the projection step as program synthesis via imitation learning, and exploit contemporary combinatorial methods for this task. We present theoretical convergence results for PROPEL and empirically evaluate the approach in three continuous control domains. The experiments show that PROPEL can significantly outperform state-of-the-art approaches for learning programmatic policies.