Say, there are two camera feeds, how can I establish if they were filming the same scene? It seems plausible that there are algorithms that somehow calculate mutual information and detect "causality violations", determining that the other feed is most probably not another perspective of the same source, and that in the presence of noise.
If it is unknown what kind of properties (e.g. dimensionality, physics) the environment that the data feeds capture has, are there algorithms that can construct the transition logic from scratch?
In computer vision, this problem is apparently called 'image registration'.