Research map / Probabilistic and scientific modeling / Molecules and proteins
Disturbance-based Discretization, Differentiable IDS Channel, and an IDS-Correcting Code for DNA-based Storage
Alan J.X. Guo, Mengyi Wei, Yufan Dai, Yali Wei, Pengchen Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 on the matching channel instead of mismatched channel (Asc test) (ablation) | THEA-Code · Asc channel | NER: 2.09 / 1.72 -> 0.90 (%) | Improved |
| train on MemSim instead of simpler C253 or C111 channel (ablation) | THEA-Code · MemSim channel, code rate 0.33 | NER: 15.9 / 17.3 -> 1.55 (%) | Improved |
| increase code rate from 0.33 to 0.83 (ablation) | THEA-Code · default 1% IDS channel | NER: 0.12 -> 3.71 (% NER, Table 1) | Hurt |
| THEA-Code vs HEDGES at low code rate 0.50 | THEA-Code · default 1% IDS channel | NER: 0.25 (HEDGES) vs 0.46 (THEA-Code) | Hurt |
| THEA-Code vs HEDGES at code rate 0.75 | THEA-Code · default 1% IDS channel | NER: 3.43 -> 2.03 | Improved |
| THEA-Code vs Cai at code rate 0.83 | THEA-Code · default 1% IDS channel | NER: 8.65 -> 2.81 | Improved |
| THEA-Code vs HEDGES on MemSim | THEA-Code · MemSim, code rate 0.50 | NER: 99.56 -> 6.11 | Improved |
| THEA-Code vs HEDGES on C111 | THEA-Code · C111 channel, code rate 0.33 | NER: 3.21 -> 2.28 | Improved |
| train on channel with higher error probability, test on lower (ablation) | THEA-Code · IDS channels 0.5%-16% | NER | Mixed |
Closest papers by meaning
- DoDo-Code: an Efficient Levenshtein Distance Embedding-based Code for 4-ary IDS Channel (NeurIPS 2025 · similarity 0.83)
- Levenshtein Distance Embedding with Poisson Regression for DNA Storage (AAAI 2024 · similarity 0.80)
- SpecMER: Fast Protein Generation with K-mer Guided Speculative Decoding (NeurIPS 2025 · similarity 0.80)
- DART-Eval: A Comprehensive DNA Language Model Evaluation Benchmark on Regulatory DNA (NeurIPS 2024 · similarity 0.80)
- Reward-Guided Iterative Refinement in Diffusion Models at Test-Time with Applications to Protein and DNA Design (ICML 2025 · similarity 0.80)
- pCoMole: Pareto-Constrained Molecule Editing with Discrete Flows (NeurIPS 2026 · similarity 0.80)
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.
