Reinforcement Learning is not a Causal problem
arXiv:1908.07617v22 citations
Originality Incremental advance
AI Analysis
This is a foundational critique for the RL community, challenging its theoretical underpinnings.
The paper argues that reinforcement learning, as currently formulated, is not a causal problem, using an analogy between non-isomorphic mathematical structures and associative versus causal information levels.
We use an analogy between non-isomorphic mathematical structures defined over the same set and the algebras induced by associative and causal levels of information in order to argue that Reinforcement Learning, in its current formulation, is not a causal problem, independently if the motivation behind it has to do with an agent taking actions.