A homotopy-type-theoretic generalization of neurosymbolic inference

In three linesTheoretical paper generalizing neurosymbolic systems using homotopy type theory. The framework preserves symmetry and multiple-proof information, converting classical functionals into belief-weighted homotopy cardinalities. Validated on MNIST reasoning-shortcut benchmarks with better calibration than diversity-trained ensembles.

Summary generated by Claude — human-verified