On whether the results are surprising: Strong theory sometimes takes years to develop and in the meantime people13 can start to get used to the main the ideas. Tomyunderstanding, thisclaim was17 often met with skepticism, especially because it lacked a frequentist regret bound to back it up. I'll try to work37 through thetime-homogeneous analysis. Some other parts ofthestate/action space arestillpoorly understood.

However, the combinatorial number of variable subsets prevents direct application of typical RL algorithms.

Reinforcement learning (RL) algorithms often require a large number of data samples to learn a control policy. As a result, training them directly on the real-world systems is expensive and potentially dangerous.

Wesummarize all unidentifiable situations that are discovered in the causal inference literature. Then, we develop a constrained optimization problem forbounding thePCfairness, whichismotivatedbythemethod proposed in[2]forbounding confounded causaleffects. Thekeyideaistoparameterize thecausal model using so-called response-function variables, whose distribution captures all randomness encoded in the causal model, so that we can explicitly traverse all possible causal models to find thetightest possible bounds.