用贝叶斯方法识别跨环境稳定的预测特征，提升模型泛化能力
Bayesian Invariance Modeling of Multi-Environment Data
- 将不变特征建模为潜在变量，通过后验推断找出稳定预测因子
- 在模拟与真实数据中，准确率和可扩展性优于现有方法
- 适合需要因果推断或跨环境泛化的研究者使用
不变预测（Invariant Prediction）[Peters et al., 2016] 分析来自多个环境的特征/结果数据，以识别具有稳定预测关系的不变特征——这些特征支持新环境下的泛化并揭示因果机制。以往方法主要依赖假设检验或正则化优化。本文提出贝叶斯不变预测（BIP），一种用于不变预测的概率模型。BIP 将不变特征的索引编码为潜变量，并通过后验推断恢复它们。在 Peters 等人 [2016] 的假设下，BIP 后验收敛于真实不变特征。我们证明了后验一致性，且环境异质性越高，后验收缩越快。为处理高维特征，设计了高效的变分近似方法 VI-BIP。在模拟与真实数据中，BIP 和 VI-BIP 均表现出更高的准确性与可扩展性，优于现有方法。
原文摘要 · Abstract (English)
Invariant prediction [Peters et al., 2016] analyzes feature/outcome data from multiple environments to identify invariant features - those with a stable predictive relationship to the outcome. Such features support generalization to new environments and help reveal causal mechanisms. Previous methods have primarily tackled this problem through hypothesis testing or regularized optimization. Here we develop Bayesian Invariant Prediction (BIP), a probabilistic model for invariant prediction. BIP encodes the indices of invariant features as a latent variable and recover them by posterior inference. Under the assumptions of Peters et al. [2016], the BIP posterior targets the true invariant features. We prove that the posterior is consistent and that greater environment heterogeneity leads to faster posterior contraction. To handle many features, we design an efficient variational approximation called VI-BIP. In simulations and real data, we find that BIP and VI-BIP are more accurate and scalable than existing methods for invariant prediction.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。