An Inside Look at Deep Neural Networks Using Graph Signal Processing

An Inside Look at Deep Neural Networks Using Graph Signal Processing
复制标题

使用图信号处理深入了解深度神经网络

DOI:
10.1109/ita.2018.8503214
复制
发表时间:
2018
期刊:
2018 Information Theory and Applications Workshop (ITA)
影响因子:
--
通讯作者:
Benjamin Girault
Benjamin Girault
中科院分区:
--
文献类型:
--
作者:
Vincent Gripon;Antonio Ortega;Benjamin Girault

文献摘要

被引文献

相似文献

深度神经网络(DNN)是许多机器学习基准测试中最先进的。了解他们的表现是一个主要的开放问题。在本文中,我们感兴趣的是使用图形信号处理来监控在一个简单的DNN架构中获得的中间表示。我们比较了不同的指标和措施,并表明,平滑的标签信号的k-最近邻图是一个很好的候选人,以解释个别层的作用,实现良好的性能。
Deep Neural Networks (DNNs) are state-of-the-art in many machine learning benchmarks. Understanding how they perform is a major open question. In this paper, we are interested in using graph signal processing to monitor the intermediate representations obtained in a simple DNN architecture. We compare different metrics and measures and show that smoothness of label signals on k-nearest neighbor graphs are a good candidate to interpret individual layers role in achieving good performance.