NeSyCat Torch: A Differentiable Tensor Implementation of Categorical Semantics for Neurosymbolic Learning

arXiv·low signal

Unifies fragmented neurosymbolic semantics (classical, fuzzy, probabilistic, neural) under a single differentiable tensor implementation grounded in categorical semantics. Theoretically ambitious but niche, with limited near-term practitioner applicability. By Daniel Romero Schellhorn, Till Mossakowski, Björn Gehrke et al. (cs.AI/cs.LG/cs.LO/math.CT).