Analysis of Deep Neural Networks with the Extended Data Jacobian Matrix

登录免费注册

资源分类

论文
算法
数据集
经验分享
技术动态
行业动态

论文
学习
研究领域

算法
学习
研究领域

数据集
自动驾驶
图片

经验分享
学习
研究领域

技术动态
计算机视觉
自然语言处理

行业动态
教育
语音识别

》资源》论文》Analysis of Deep Neural Networks with the Extended Data Jacobian Matrix

Analysis of Deep Neural Networks with the Extended Data Jacobian Matrix

2020-03-06 |

61 |

33 |

Analysis of Deep Neural Networks with the Extended Data Jacobian Matrix
论文

Abstract

Deep neural networks have achieved great success on a variety of machine learning tasks. There are many fundamental and open questions yet to be answered, however. We introduce the Extended Data Jacobian Matrix (EDJM) as an architecture-independent tool to analyze neural networks at the manifold of interest. The spectrum of the EDJM is found to be highly correlated with the complexity of the learned functions. After studying the effect of dropout, ensembles, and model distillation using EDJM, we propose a novel spectral regularization method, which improves network performance.

上一篇：Barron and Cover’s Theory in Supervised Learning and its Application to Lasso

下一篇：On collapsed representation of hierarchical Completely Random Measures

用户评价