arXiv cs.AI· Tinglan Wang, Qun Su, Shihang Wang, Chi Wang, Lin Wang, Shaobo Chen, Haixu Liu, Hai Wei, Yin Ma, Yu Kang, Qi Gao, Xiaojun Yao, Tingjun Hou, Kai Wen·· 1 小时前AI 评分33
Structured Re-Decoding（SRD）：无需训练扩展预训练模型的决策空间
Expanding the Decision Space of Pretrained Models through Structured Re-Decoding
AI 导读
arXiv 论文提出 Structured Re-Decoding（SRD），一种免训练框架，通过构建本地候选并联合优化其选择、验证重建输出来突破原生解码器的限制。在分子生成、蛋白质设计、RNA 结构预测和抗体建模等任务中，SRD 改善了结构化推理，可产生原样本池中不存在的已验证分子。外部化学模板还能扩展到非规范氨基酸与肽环化，物理相干 Ising 机上的初步实验验证了硬件优化的可行性。
正文 · 原文
Authors:Tinglan Wang, Qun Su, Shihang Wang, Chi Wang, Lin Wang, Shaobo Chen, Haixu Liu, Hai Wei, Yin Ma, Yu Kang, Qi Gao, Xiaojun Yao, Tingjun Hou, Kai Wen
Abstract:Native decoders limit how pretrained models combine local predictions under new structural constraints. We introduce Structured Re-Decoding (SRD), a training-free framework that constructs local candidates from model scores, samples, or external specifications, jointly optimizes their selection, and verifies reconstructed outputs. We show that candidate recombination expands feasible support and can improve attainable reward under a fixed KL divergence budget. Across molecular generation, protein design, RNA structure prediction, and antibody modeling, SRD enables improved structured inference, including verified molecules absent from the original sample pool. External chemical templates further extend SRD to noncanonical amino acids, peptide cyclization, and grammar-guided molecular construction. Preliminary experiments on a physical coherent Ising machine demonstrate the feasibility of hardware-based optimization, suggesting broader opportunities for quantum and quantum-inspired approaches to structured inference and combinatorial optimization across scientific domains.
| Comments: | 29 pages, 2 figures |
| Subjects: | Artificial Intelligence (cs.AI); Biomolecules (q-bio.BM) |
| Cite as: | arXiv:2611.15109 [cs.AI] |
| (or arXiv:2611.15109v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.15109 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tinglan Wang [view email]
[v1]
Wed, 11 Nov 2026 06:37:08 UTC (1,212 KB)
来源：arXiv cs.AI · arxiv.org