Influence functions (IFs) are a powerful tool for detecting anomalous examples in large scale datasets. However, they are unstable when applied to deep networks. In this paper, we provide an explanation for the instability of IFs and develop a solution to this problem. We show that IFs are unreliable when the two data points belong to two different classes. Our solution leverages class information to improve the stability of IFs. Extensive experiments show that our modification significantly improves the performance and stability of IFs while incurring no additional computational cost.
翻译:影响函数(IFs)是大规模数据集中检测异常样本的强大工具。然而,它们在应用于深层网络时是不稳定的。在这篇论文中,我们提供了IFs不稳定性的解释并开发了解决此问题的方案。我们发现,当两个数据点属于不同的类时,IFs是不可靠的。我们的解决方案利用类别信息来提高IFs的稳定性。广泛的实验表明,我们的修改显著提高了IFs的性能和稳定性,而不会产生额外的计算成本。