arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1901.08557cs.LGstat.ML
收藏

网络科学与互信息对深层神经网络的解释

On Network Science and Mutual Information for Explaining Deep Neural Networks

Carnegie Mellon University(卡内基梅隆大学);University of Cambridge(剑桥大学);Arm Inc.(安谋公司);The University of Texas at Austin(德克萨斯大学奥斯汀分校)

Brian Davis, Umang Bhatt, Kartikeya Bhardwaj, Radu Marculescu, José M. F. Moura

摘要:

在本文中,我们提出了一种解释深度学习模型的新方法。通过将互信息与网络科学相结合,我们探索了信息如何在前馈网络中流动。我们表明,有效地近似互信息使我们能够创建一个信息度量,量化深度学习模型的任何两个神经元之间有多少信息流。为此,我们提出了NIF,神经信息流,一种用于编码信息流的技术,它暴露了深度学习模型内部并提供了特征属性。

英文摘要:

In this paper, we present a new approach to interpret deep learning models. By coupling mutual information with network science, we explore how information flows through feedforward networks. We show that efficiently approximating mutual information allows us to create an information measure that quantifies how much information flows between any two neurons of a deep learning model. To that end, we propose NIF, Neural Information Flow, a technique for codifying information flow that exposes deep learning model internals and provides feature attributions.

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。

↑