Logic-Guided Vector Fields for Constrained Generative Modeling

Ali Baheri · 2026

Read Logic-Guided Vector Fields for Constrained Generative Modeling by Ali Baheri (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-02009.txt.

Description

Ali Baheri proposes Logic-Guided Vector Fields, a neuro symbolic frame injecting constraints into flow models. Two mechanisms combine: a training time logic loss penalizing violations along paths, weighting near target; and an inference time adjustment steering sampling with gradients as a light correction. Three cases spanning linear, nonlinear and multi region feasibility show violations falling fifty nine to eighty two percent. In linear and ring settings, fidelity grows, while multi obstacle trades feasibility for fit. The framework also yields constraint aware vector fields showing obstacle dodging.

Catalog record

CREATORAli Baheri
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40051
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.02009

Browse the full library · About the Tower of Babel