Research map / Probabilistic and scientific modeling / Variational and generative sampling
Where to Approximate in Neurosymbolic Inference?
Samy Badreddine, Emile van Krieken, Luciano Serafini, Antonio Vergari
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Unifies approximate weighted model counting for neurosymbolic inference under a three-step bottom-up approximate compilation pipeline and studies where in that pipeline the approximation budget is best spent.
Problem
Exact weighted model counting (WMC), which underlies probabilistic neurosymbolic methods, is #P-hard and scales poorly, so approximations are needed; it is unclear where in the computation the approximation budget should be allocated.
Approach
Unifies existing approximate WMC approaches as a three-step bottom-up approximate compilation pipeline and uses this lens to analyze where approximation budget is best allocated, drawing inspiration from tensor network approximation methods.
Closest papers by meaning
- On the Hardness of Probabilistic Neurosymbolic Learning (ICML 2024 · similarity 0.91)
- CTSketch: Compositional Tensor Sketching for Scalable Neurosymbolic Learning (NeurIPS 2025 · similarity 0.88)
- On the Independence Assumption in Neurosymbolic Learning (ICML 2024 · similarity 0.87)
- A Fast Convoluted Story: Scaling Probabilistic Inference for Integer Arithmetics (NeurIPS 2024 · similarity 0.86)
- DOLPHIN: A Programmable Framework for Scalable Neurosymbolic Learning (ICML 2025 · similarity 0.85)
- EM-NeSy: Expectation Maximization for Neurosymbolic Learning (NeurIPS 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
