arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2012.10425cs.LG
收藏

深度神经网络的鲁棒性分析

Towards Robust Explanations for Deep Neural Networks

Technische Universität Berlin(柏林工业大学);BIFOLD - Berlin Institute for the Foundations of Learning and Data(BIFOLD——柏林学习与数据基础研究所);Korea University(高丽大学);Max Planck Institute for Informatics(马克斯·普朗克信息学研究所);Google Research(谷歌研究院)

Ann-Kathrin Dombrowski, Christopher J. Anders, Klaus-Robert Müller, Pan Kessel

摘要:

解释方法揭示了黑盒分类器(如深度神经网络)的决策过程。但它们的有用性可能会受到损害,因为它们容易受到操纵。通过这项工作,我们的目标是增强解释的弹性。我们开发了一个统一的理论框架,推导出一个模型的最大可操作性的界限。基于这些理论见解,我们提出了三种不同的技术来提高对操纵的鲁棒性:用权重衰减训练,平滑激活函数和最小化网络的Hessian。我们的实验结果证实了这些方法的有效性。

英文摘要:

Explanation methods shed light on the decision process of black-box classifiers such as deep neural networks. But their usefulness can be compromised because they are susceptible to manipulations. With this work, we aim to enhance the resilience of explanations. We develop a unified theoretical framework for deriving bounds on the maximal manipulability of a model. Based on these theoretical insights, we present three different techniques to boost robustness against manipulation: training with weight decay, smoothing activation functions, and minimizing the Hessian of the network. Our experimental results confirm the effectiveness of these approaches.

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。

↑