Research map / Probabilistic and scientific modeling / Variational and generative sampling
DeepStochLog: Neural Stochastic Logic Programming
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use stochastic logic programming with neural grammar rules rather than possible-world neural logic programming | DeepStochLog vs DeepProbLog · MNIST Addition, 4 digits | inference time (ms): 5.7 ± 1.8; timeout | Helped |
| Use DeepStochLog instead of DeepProbLog | MNIST Addition · 3 digits | accuracy (%): 94.5 ± 1.1; timeout | Mixed |
| Use DeepStochLog instead of DeepProbLog | HWF · expression length 7 | accuracy (%): 94.8 ± 0.9; 20.4 ± 37.2 | Helped |
| Use DeepStochLog instead of DeepProbLog | Well-formed Parentheses · maximum expression length 18 | parse accuracy (%): 99.2 ± 0.8 -> 100.0 ± 0.0 | Helped |
| Use DeepStochLog instead of DeepProbLog | a^n b^n c^n recognition · expression length 3–12 | accuracy (%): 99.8 ± 0.3 -> 99.4 ± 0.5 | Hurt |
| Use DeepStochLog rather than task-specific GCN features | Citation classification · Cora | accuracy (%): 81.5 -> 69.4 | Hurt |
| Use DeepStochLog rather than task-specific GCN features | Citation classification · Citeseer | accuracy (%): 70.3 -> 65.0 | Hurt |
| Remove tabling from DeepStochLog inference (ablation) | DeepStochLog · HWF, expression length 5 | parsing time (seconds): 0.95 -> 3.78 | Hurt |
| Remove tabling from DeepStochLog inference (ablation) | DeepStochLog · HWF, expression length 7 | parsing time (seconds): 10.95 -> 30.42 | Hurt |
| Remove tabling from DeepStochLog inference (ablation) | DeepStochLog · HWF, expression length 1 | parsing time (seconds): 0.060 -> 0.067 | No clear effect |
| Remove tabling from DeepStochLog inference (ablation) | DeepStochLog · HWF, expression length 3 | parsing time (seconds): 0.096 -> 0.081 | Helped |
| Use DeepStochLog instead of DeepProbLog | Word Algebra Problems · Word Algebra Problems | accuracy (%): 94.2 ± 1.4 -> 94.8 ± 1.1 | No clear effect |
Closest papers by meaning
- DeepProbLog: Neural Probabilistic Logic Programming (NeurIPS 2018 · similarity 0.93)
- Soft-Unification in Deep Probabilistic Logic (NeurIPS 2023 · similarity 0.88)
- DeepProofLog: Efficient Proving in Deep Stochastic Logic Programs (AAAI 2026 · similarity 0.88)
- Neuro-Symbolic Temporal Point Processes (ICML 2024 · similarity 0.85)
- LogicMP: A Neuro-symbolic Approach for Encoding First-order Logic Constraints (ICLR 2024 · similarity 0.85)
- On the Hardness of Probabilistic Neurosymbolic Learning (ICML 2024 · similarity 0.84)
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.