Combined Reinforcement Learning via Abstract Representations
This approach addresses the problem of robust and efficient reinforcement learning for AI systems, offering potential benefits in interpretability, exploration, and transfer learning, though it appears incremental as it builds on existing paradigms.
The paper tackles the challenge of combining model-free and model-based reinforcement learning by introducing a shared low-dimensional learned encoding of the environment, which improves generalization and computational efficiency through planning in a smaller latent space.
In the quest for efficient and robust reinforcement learning methods, both model-free and model-based approaches offer advantages. In this paper we propose a new way of explicitly bridging both approaches via a shared low-dimensional learned encoding of the environment, meant to capture summarizing abstractions. We show that the modularity brought by this approach leads to good generalization while being computationally efficient, with planning happening in a smaller latent state space. In addition, this approach recovers a sufficient low-dimensional representation of the environment, which opens up new strategies for interpretable AI, exploration and transfer learning.