Research map / Learning theory and algorithms / Combinatorial and search optimization
Constraining Linear-chain CRFs to Regular Languages
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 |
|---|---|---|---|
| Train with regular-language constraints instead of applying them only at decoding | RegCCRF with RoBERTa · OntoNotes evaluation | F1: 87.21 -> 87.53 | Helped |
| Train with regular-language constraints instead of applying them only at decoding | RegCCRF with RoBERTa · OntoNotes evaluation | Recall: 87.38 -> 87.76 | Helped |
| Apply constraints at decoding | RoBERTa CRF · OntoNotes evaluation | Precision: 86.92 -> 87.05 | Helped |
| Apply constraints at decoding rather than decode without constraints | Reduced-role RoBERTa CRF · OntoNotes evaluation | Recall: 87.35 -> 87.38 | No clear effect |
| Train with constraints rather than train unconstrained and constrain decoding | Synthetic CRF with language (ac)* | (bc)* · Synthetic sequences of length 2k | Negative log-likelihood | Helped |
| Train with constraints rather than train unconstrained and constrain decoding | Synthetic CRF with language acd | bcd | bce · MAP inference | Probability assigned to data-distribution mode acd: 0.32 -> 0.40 | Helped |
| Train unconstrained, then apply constraints at decoding | Synthetic CRF with language acd | bcd | bce · MAP inference | Probability assigned to bcd: 0.30 -> 0.48 | Hurt |
Closest papers by meaning
- Frame Semantic Role Labeling Using Arbitrary-Order Conditional Random Fields (AAAI 2024 · similarity 0.87)
- Constrained Decoding for Diffusion Language Models via Efficient Inference over Finite Automata (NeurIPS 2026 · similarity 0.85)
- Chain-of-Model Learning for Language Model (NeurIPS 2025 · similarity 0.84)
- Dynamically Structured Diffusion Language Model Decoding via Bayesian Inference (NeurIPS 2026 · similarity 0.84)
- Short-Context Dominance: How Much Local Context Natural Language Actually Needs? (NeurIPS 2026 · similarity 0.84)
- Reconsidering Positional Supervision in Masked Diffusion Language Model Training (NeurIPS 2026 · similarity 0.83)
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.