通过欺骗分布偏移检测器，让模型学会识别稳定特征以提升跨域泛化能力。
Deceptive Risk Minimization: Out-of-Distribution Generalization by Deceiving Distribution Shift Detectors
- 用可微目标同时欺骗检测器并优化任务损失，实现隐式分布对齐。
- 在概念漂移和协变量漂移场景中均显著提升未见领域表现。
- 无需测试数据或预设域划分，适合真实部署环境中的鲁棒学习。
本文提出一种基于欺骗的分布外（OOD）泛化机制：通过学习使训练数据对观察者呈现独立同分布（iid）特性的数据表示，识别出稳定特征以消除虚假相关性，从而实现对未知领域的泛化。该原则称为欺骗风险最小化（DRM），并设计了一个可微的目标函数，同时学习使基于共形马尔可夫链的检测器无法察觉分布偏移的特征，并最小化特定任务损失。与域自适应或已有不变表示学习方法不同，DRM无需测试数据或预先划分训练数据为有限数量的生成域。我们在数值实验中验证了概念漂移以及模拟机器人模仿学习场景下的协变量漂移设置下，DRM的有效性。
原文摘要 · Abstract (English)
This paper proposes deception as a mechanism for out-of-distribution (OOD) generalization: by learning data representations that make training data appear independent and identically distributed (iid) to an observer, we can identify stable features that eliminate spurious correlations and generalize to unseen domains. We refer to this principle as deceptive risk minimization (DRM) and instantiate it with a practical differentiable objective that simultaneously learns features that eliminate distribution shifts from the perspective of a detector based on conformal martingales while minimizing a task-specific loss. In contrast to domain adaptation or prior invariant representation learning methods, DRM does not require access to test data or a partitioning of training data into a finite number of data-generating domains. We demonstrate the efficacy of DRM on numerical experiments with concept shift and a simulated imitation learning setting with covariate shift in environments that a robot is deployed in.
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。