arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.09344cs.LGcs.AIcs.CV
收藏

超越分布偏移:通过训练动态的视角看虚假特征

Beyond Distribution Shift: Spurious Features Through the Lens of Training Dynamics

University of Pittsburgh(匹兹堡大学);New York University(纽约大学);Boston University(波士顿大学)

Nihal Murali, Aahlad Puli, Ke Yu, Rajesh Ranganath, Kayhan Batmanghelich

AI总结:

本文通过训练动态视角区分良性/有害虚假特征,提出用实例难度指标(如Prediction Depth)在早期训练中检测有害特征,并验证其有效性,强调需监控训练动态而非仅准确性。

摘要:

深度神经网络(DNN)在训练过程中容易学习到与标签相关但与学习问题无关的虚假特征。这损害了模型的泛化能力,并在安全关键应用中部署时带来问题。本文旨在通过训练过程中内部神经元的学习动态的视角,更好地理解虚假特征的影响。我们做出以下观察:(1)虽然先前的工作强调了虚假特征对DNN泛化能力的有害影响,但我们强调并非所有虚假特征都是有害的。虚假特征可以是“良性的”或“有害的”,这取决于对于给定模型,它们是否比核心特征“更难”或“更容易”学习。这个定义依赖于模型和数据集。(2)我们基于这一前提,使用实例难度方法(如Prediction Depth(Baldock等人,2021))来量化给定模型的“易学性”,并在训练阶段识别这种行为。(3)我们通过实验表明,可以通过观察DNN早期层的学习动态来检测有害的虚假特征。换句话说,在训练早期由DNN的初始层学习的容易特征可能(潜在地)损害模型的泛化能力。我们在医学和视觉数据集(包括模拟和真实数据)上验证了我们的主张,并通过展示Prediction Depth与信息论概念(如V-usable information(Ethayarajh等人,2021))之间的理论联系,证明了我们假设的经验成功。最后,我们的实验表明,仅监控训练过程中的准确性(这在机器学习流程中很常见)不足以检测虚假特征。因此,我们强调需要使用合适的实例难度指标来监控早期训练动态。

英文摘要:

Deep Neural Networks (DNNs) are prone to learning spurious features that correlate with the label during training but are irrelevant to the learning problem. This hurts model generalization and poses problems when deploying them in safety-critical applications. This paper aims to better understand the effects of spurious features through the lens of the learning dynamics of the internal neurons during the training process. We make the following observations: (1) While previous works highlight the harmful effects of spurious features on the generalization ability of DNNs, we emphasize that not all spurious features are harmful. Spurious features can be "benign" or "harmful" depending on whether they are "harder" or "easier" to learn than the core features for a given model. This definition is model and dataset-dependent. (2) We build upon this premise and use instance difficulty methods (like Prediction Depth (Baldock et al., 2021)) to quantify "easiness" for a given model and to identify this behavior during the training phase. (3) We empirically show that the harmful spurious features can be detected by observing the learning dynamics of the DNN's early layers. In other words, easy features learned by the initial layers of a DNN early during the training can (potentially) hurt model generalization. We verify our claims on medical and vision datasets, both simulated and real, and justify the empirical success of our hypothesis by showing the theoretical connections between Prediction Depth and information-theoretic concepts like V-usable information (Ethayarajh et al., 2021). Lastly, our experiments show that monitoring only accuracy during training (as is common in machine learning pipelines) is insufficient to detect spurious features. We, therefore, highlight the need for monitoring early training dynamics using suitable instance difficulty metrics.

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。

↑