arXiv cs.AI· Mohamed Bouadi, Nassim Bouarour, Varun Kulkarni, Shivam Dubey, Aditya Tanna, Vinay Kumar Sankarapu·· 12 小时前AI 评分32
O'Prior：合成任务分布如何决定表格基础模型质量
Shaping the Prior: How Synthetic Task Distributions Determine Tabular Foundation Model Quality
AI 导读
研究提出 O'Prior 合成先验，通过分层 SCM 元生成器、模块化真实感引擎、显式压力模块和课程化防泄漏生成协议四个组件构建表格基础模型的预训练分布。在固定架构、优化器和算力预算、仅改变合成任务分布的条件下，O'Prior 在真实表格基准上取得一致且显著的下游准确率与鲁棒性提升，增益集中在分布不规则的场景。消融实验显示机制多样性、真实感组合与偏移感知压力各自独立贡献，不可互相替代。
正文 · 原文
Abstract:What determines the quality of a tabular foundation model? Unlike language or vision, tabular foundation models acquire their inductive biases almost entirely from synthetic pretraining distributions, yet the design of these distributions remains poorly understood. Standard synthetic priors are too well-behaved: they omit the irregularities and failure modes that determine deployment robustness. We introduce O'Prior, a compositional realism prior built around four coupled components: a hierarchical SCM meta-generator spanning diverse functional families; a modular realism engine covering heterogeneous marginals, missingness, and target transforms; an explicit stress module injecting confounding and support-query mismatch; and a curriculum-governed, leakage-safe generation protocol. To isolate prior design as the scientific variable, we hold architecture, optimizer, and compute budget fixed and vary only the synthetic task distribution. O'Prior yields consistent and substantial improvements in downstream accuracy and robustness across real tabular benchmarks, with gains concentrated in regimes characterized by distributional irregularities. Ablations confirm that mechanism diversity, realism composition, and shift-aware stress each contribute independently, their effects are not interchangeable. These results establish synthetic prior construction as a first-order and largely overlooked determinant of tabular foundation model quality.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.18971 [cs.LG] |
| (or arXiv:2605.18971v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2605.18971 arXiv-issued DOI via DataCite |
Submission history
From: Mohamed Bouadi [view email]
[v1]
Mon, 18 May 2026 18:00:45 UTC (4,988 KB)
[v2]
Fri, 23 Oct 2026 12:04:54 UTC (4,988 KB)
来源：arXiv cs.AI · arxiv.org