arXiv：cs.LG· Xuan Gong, Hanbo Huang, Wenbin Dai, Jing Wang, Lei Bai, Xiang Xiao, Weishu Zhao, Shiyu Liang·· 4 小时前
QuotientPO：面向基因组尺度代谢模型修复的商空间探索
Beyond Action Entropy: Quotient-Space Exploration for Genome-Scale Metabolic Model Repair
AI 导读
QuotientPO 将等价修复折叠为规范机制，并直接在商空间上优化探索，在 2,212 个留出 GEM 上将 Success@32 从 17.93% 提升至 20.10%（相对提升 12.1%），同时在相同采样预算下持续发现更多不同的成功修复核心。
正文
Abstract:Repairing scientific models from functional observations differs fundamentally from supervised prediction: feedback may certify a solution without revealing which structural correction is responsible. We study this setting for genome-scale metabolic model (GEM) repair, where multiple reaction edits can explain the same phenotypes and many apparently distinct edits correspond to the same biological mechanism. This many-to-one structure creates a hidden failure mode for conventional exploration: diversity in the output space need not translate into diversity of scientific hypotheses. We introduce QuotientPO, which collapses equivalent repairs into canonical mechanisms and optimizes exploration directly over the resulting quotient space. To make quotient exploration informative under finite rollouts, we derive a kernelized Rényi estimator that resolves graded crowding among distinct repair cores beyond coarse exact-match counts. On 2,212 held-out GEMs, QuotientPO improves Success@32 from 17.93% to 20.10% (+12.1% relative) while consistently increasing distinct successful-core discovery under the same sampling budget. These results establish quotient-space exploration as a principled approach to mechanism-level discovery under verifier-induced equivalence.
| Comments: | Under Review |
| Subjects: | Machine Learning (cs.LG); Molecular Networks (q-bio.MN) |
| Cite as: | arXiv:2610.11627 [cs.LG] |
| (or arXiv:2610.11627v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11627 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xuan Gong [view email]
[v1]
Thu, 8 Oct 2026 10:05:57 UTC (30,687 KB)
来源：arXiv：cs.LG · arxiv.org