Approximate Bayesian inversions in probabilistic quasi-Borel spaces

Let QBS\mathbf{QBS} be the category of quasi-Borel spaces, and let P:QBSQBS\mathbb{P}:\mathbf{QBS}\to\mathbf{QBS} denote the probabilistic construction. For objects AA and BB, a morphism f:QBS(A,PB)f:\mathbf{QBS}(A,\mathbb{P}B) represents a probabilistic map from AA to BB. Consider an inference functor FF^{\dagger} whose action on objects is the identity and whose action on morphisms is similar to FqF_q above. Approximate Bayesian inversions in probabilistic quasi-Borel spaces. Such a functor should be a dagger endofunctor

F:P:QBSP:QBSF^{\dagger}:\mathbb{P}:\mathbf{QBS}\rightarrow\mathbb{P}:\mathbf{QBS}

sending objects to themselves and morphisms f:QBS(A,PB)f:\mathbf{QBS}(A,\mathbb{P}B) to approximate Bayesian inversions

f:QBS(B,PA).f^{\dagger}:\mathbf{QBS}(B,\mathbb{P}A).

This would extend the paper's inference construction beyond faithful inverses for string diagrams and provide an approximation to Bayesian inversion in the sense of Cho and Jacobs. Whether the category of quasi-Borel spaces has the conditionals needed to support such a dagger functor is not yet known.

Sources & referencesView supporting material

Primary source

Eli Sennesh, Tom Xu and Yoshihiro Maruyama, “A Probabilistic Generative Model of Free Categories”, arXiv:2205.04545 (2022).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.