arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1906.04868cs.LGstat.ML
收藏

嵌入神经网络过参数化的半平坦极小点和鞍点

Semi-flat minima and saddle points by embedding neural networks to overparameterization

The Institute of Statistical Mathematics(统计数理研究所);Preferred Networks, Inc.(Preferred Networks公司);RIKEN(理化学研究所)

Kenji Fukumizu, Shoichiro Yamaguchi, Yoh-ichi Mototake, Mirai Tanaka

摘要:

我们从理论上研究了过参数化情况下神经网络的训练误差。我们考虑了三种基本的方法嵌入一个网络到一个更广泛的一个更多的隐藏单元,并讨论是否最小点的较窄的网络给出了最小值或鞍点的较宽的。我们的研究结果表明,具有平滑和ReLU激活的网络在嵌入点周围具有不同的部分平坦景观。我们还将这些结果的差异,他们的泛化能力overparameterized实现。

英文摘要:

We theoretically study the landscape of the training error for neural networks in overparameterized cases. We consider three basic methods for embedding a network into a wider one with more hidden units, and discuss whether a minimum point of the narrower network gives a minimum or saddle point of the wider one. Our results show that the networks with smooth and ReLU activation have different partially flat landscapes around the embedded point. We also relate these results to a difference of their generalization abilities in overparameterized realization.

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。

↑