Scalable Causal Domain Adaptation
Mohammad Ali Javidian School of Electrical and Computer Engineering Purdue University、Om Pandey School of Computer Engineering Kalinga Institute of Industrial Technology Pooyan Jamshidi Computer Science、Engineering Department University of South Carolina
Abstract
One of the most critical problems in transfer learning is the task of domain adaptation, where the goal is to apply an algorithm trained in one or more source domains to a different (but related) target domain. This paper deals with domain adaptation in the presence of covariate shift while invariance exist across domains. One of the main limitations of existing causal inference methods for solving this problem is scalability. To overcome this difficulty, we propose SCTL, an algorithm that avoids an exhaustive search and identifies invariant causal features across source and target domains based on Markov blanket discovery. SCTL does not require having prior knowledge of the causal structure, the type of interventions, or the intervention targets. There is an intrinsic locality associated with SCTL that makes it practically scalable and robust because local causal discovery increases the power of computational independence tests and makes the task of domain adaptation computationally tractable. We show the scalability and robustness of SCTL for domain adaptation using synthetic and real data sets in low-dimensional and high-dimensional settings.
中文速览
在跨域预测任务中，训练数据和测试数据往往来自不同分布，如何找到在各域间保持稳定的因果特征是迁移学习的核心难题。现有因果推断方法虽能识别这类不变特征，但面对高维数据时需要指数级数量的条件独立性检验，几乎无法扩展到几十个变量以上。为此，作者提出了SCTL（可扩展因果迁移学习）算法，核心思路是把搜索范围从所有变量收窄到目标变量的马尔可夫毯（Markov blanket）——由于马尔可夫毯在实际因果模型中通常只有不到10个变量，检验次数大幅下降，独立性测试的统计效力也随之提升，整个算法在高维场景下依然可行。作者还从理论上证明，即便存在未观测混杂因素，主流马尔可夫毯发现算法（如GSMB、IAMB）在忠实性假设下仍然正确，从而为SCTL提供了严格的理论保障。在合成数据和真实数据（包括糖尿病早期预测等任务）上的实验表明，SCTL在协变量偏移场景下的预测精度优于多个现有方法，为高维领域自适应提供了兼具理论支撑与实用价值的新途径。
原文 arXiv:2103.00139;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2103.00139v3