arXiv：cs.LG· Mohamed Bouadi, Nassim Bouarour, Shivam Dubey, Aditya Tanna, Vinay Kumar Sankarapu·· 3 小时前AI 评分33
从行为到溯源：将表格基础模型归因于合成预训练数据
From Behavior to Provenance: Attributing Tabular Foundation Models to Synthetic Pretraining Data
AI 导读
研究提出用 O'PRIOR 合成任务生成器构建可验证的归因测试平台，结合行为条件归因、反事实重训练与溯源干预来检验任务级忠实度与机制级一致性。在留出真实任务上，移除归因最高的 5% 合成任务使平均 ROC-AUC 下降 0.013，随机移除仅 0.002±0.004，移除最低归因任务则提升 0.003。溯源判别按 AUROC 排序仅为 0.55-0.62，表明溯源关联与干预忠实度并不必然一致。
正文
Abstract:Training-data attribution aims to identify which training examples shape model behavior, yet validating such claims is difficult because causal training influence is rarely observable. We argue that controlled synthetic pretraining makes attribution experimentally testable. Using O'PRIOR, a provenance-rich synthetic task generator for tabular foundation models, we construct a testbed in which every pretraining task carries explicit lineage over structural mechanisms, missingness, confounding, shortcuts, and distribution shift. We combine behavior-conditioned attribution with counterfactual retraining and provenance-aware interventions to test both task-level faithfulness and mechanism-level consistency. On held-out real tasks, removing the top-attributed 5% of synthetic tasks decreases mean ROC-AUC by 0.013, compared with 0.002$\pm$0.004 under random removal, while removing bottom-attributed tasks improves performance by 0.003. Within shortcut-provenance tasks, targeted removal yields an effect of 0.043 versus 0.016 for matched random removal. Provenance discrimination is more modest by ranking AUROC (0.55-0.62), despite substantial top-k enrichment, revealing that provenance association and interventional faithfulness need not coincide. Our results establish synthetic provenance as a controlled setting for verifiable contributive attribution
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.02347 [cs.LG] |
| (or arXiv:2610.02347v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02347 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Mohamed Bouadi [view email]
[v1]
Thu, 1 Oct 2026 18:21:32 UTC (243 KB)
来源：arXiv：cs.LG · arxiv.org