arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1801.06801cs.LGcs.CV
收藏

两种神经网络的曲率比较

Curvature-based Comparison of Two Neural Networks

Shanghai Jiao Tong University(上海交通大学)

Tao Yu, Huan Long, John E. Hopcroft

摘要:

本文通过比较两个深度神经网络的每一个全连通层中由激活向量组成的流形,展示了它们的异同。本文的主要贡献包括:1)一个新的数据生成算法,这是至关重要的确定维数的流形; 2)一个系统的战略比较流形。特别地,我们把黎曼曲率和截面曲率作为判据的一部分,它能反映流形的内在几何性质。给出了一些有趣的结果和现象,有助于说明两种网络提取的特征之间的异同,揭开深层神经网络内在机理的神秘面纱。

英文摘要:

In this paper we show the similarities and differences of two deep neural networks by comparing the manifolds composed of activation vectors in each fully connected layer of them. The main contribution of this paper includes 1) a new data generating algorithm which is crucial for determining the dimension of manifolds; 2) a systematic strategy to compare manifolds. Especially, we take Riemann curvature and sectional curvature as part of criterion, which can reflect the intrinsic geometric properties of manifolds. Some interesting results and phenomenon are given, which help in specifying the similarities and differences between the features extracted by two networks and demystifying the intrinsic mechanism of deep neural networks.

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。

↑