arXiv：cs.LG（机器学习，全量分类）· Yifan Guo·· 14 小时前AI 评分32
MECHVAR：面向自主机器学习实验选择的方差引导机制判别方法
MECHVAR: Variance-Guided Mechanism Discrimination for Autonomous Machine Learning Experiment Selection
AI 导读
MECHVAR 通过最大化候选机制预测响应的后验加权方差来选择下一个实验探针，在共享高斯预测模型下与经典 Box-Hill 后验加权成对 KL 准则成正比，并支持 O(KE) 向量化重打分。
正文
Abstract:Benchmark gains are often mechanism-ambiguous: reproducing an improvement does not by itself identify why it occurs. We study finite-library mechanism discrimination, where posterior-weighted candidate mechanisms, executable probes, and a limited experimental budget define a sequential experiment-selection problem. MECHVAR selects the next probe by maximizing the posterior-weighted variance of its predicted responses. Under a shared-Gaussian predictive model, this score is exactly proportional to the classical Box--Hill posterior-weighted pairwise-KL criterion, yet it admits O(KE) vectorized rescoring and a transparent additive audit over mechanism pairs. A local expansion further links the score to expected information gain (EIG) when predicted response separations are small. In a 25-block stress audit, MECHVAR outperforms confirmation-first in several moderate misspecification regimes, while its primary comparisons with EIG remain statistically unresolved. In a held-out Digits loop, normalized mechanism-identification AUC is 0.8975 for MECHVAR, 0.7825 for a score-greedy policy, and 0.9092 for EIG. At K = 100, E = 200, median single-thread full-library scoring is 10.36 microseconds for MECHVAR versus 57.69 ms for six-node quadrature EIG in the recorded environment. MECHVAR therefore provides a lightweight, auditable acquisition rule for finite-library experiment selection when a shared predictive scale is a defensible approximation.
| Comments: | 17 pages, 7 figures |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.01819 [cs.LG] |
| (or arXiv:2610.01819v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.01819 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yifan Guo [view email]
[v1]
Thu, 1 Oct 2026 14:53:09 UTC (705 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org