arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.18564cs.LGcs.AI
收藏

Transformer 学习在分布不确定性下的稳健上下文回归

Transformers Learn Robust In-Context Regression under Distributional Uncertainty

Faculty of Computer Science;Engineering, Ho Chi Minh University of Technology (HCMUT), Vietnam National University Ho Chi Minh City (VNU-HCM), Vietnam

Hoang T. H. Cao, Hai D. V. Trinh, Tho Quan, Lan V. Truong

AI总结:

研究探讨了Transformer在非高斯系数、重尾噪声和非i.i.d.提示下的稳健上下文回归能力,证明其在多种分布偏移下优于传统基线方法。

摘要:

近期研究表明,Transformer可以在限制性假设下执行上下文学习进行线性回归,包括i.i.d.数据、高斯噪声和高斯回归系数。然而,现实数据常违反这些假设:输入分布、噪声分布和系数分布通常未知、非高斯且可能在提示中存在依赖性。这提出了一个根本性问题:在现实分布不确定性下,Transformer能否有效进行上下文学习?我们研究了在广泛分布偏移下的嘈杂线性回归的上下文学习,包括非高斯系数、重尾噪声和非i.i.d.提示。我们比较了Transformer与在相应最大似然准则下最优或次优的传统基线方法。在所有设置中,Transformer一致匹配或优于这些基线,证明了其在传统估计器之外的稳健上下文适应能力。

英文摘要:

Recent work has shown that Transformers can perform in-context learning for linear regression under restrictive assumptions, including i.i.d. data, Gaussian noise, and Gaussian regression coefficients. However, real-world data often violate these assumptions: the distributions of inputs, noise, and coefficients are typically unknown, non-Gaussian, and may exhibit dependency across the prompt. This raises a fundamental question: can Transformers learn effectively in-context under realistic distributional uncertainty? We study in-context learning for noisy linear regression under a broad range of distributional shifts, including non-Gaussian coefficients, heavy-tailed noise, and non-i.i.d. prompts. We compare Transformers against classical baselines that are optimal or suboptimal under the corresponding maximum-likelihood criteria. Across all settings, Transformers consistently match or outperform these baselines, demonstrating robust in-context adaptation beyond classical estimators.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑