Uncertainty in category theory
What does category theory have to say about missing data, incomplete information, and other forms of uncertainty?
Literature
General
- Johnson & Rosebrugh, 2003: Three approaches to partiality in the sketch data model (doi, pdf)
- Marsden, 2016: Ambiguity and incomplete information in categorical models of
language (arxiv, pdf, slides)
- General strategy:
- Define a commutative monad \(T\) on \(\mathbf{Set}\)
- Its Eilenberg-Moore algebras form a bicomplete, closed symmetric monoidal category \(\mathrm{EM}(T)\)
- Take categories enriched in \(\mathrm{EM}(T)\), usually generated freely from an existing category like \(\mathbf{Rel}\) or \(\mathbf{FdHilb}\)
- Includes both qualitative and quantitative models
- The qualitative models are not very convincing
- General strategy:
Probabilistic
TODO