BriefGPT.xyz
Jun, 2019
缓解宽神经网络病态尖锐度的规范化方法
The Normalization Method for Alleviating Pathological Sharpness in Wide Neural Networks
HTML
PDF
Ryo Karakida, Shotaro Akaho, Shun-ichi Amari
TL;DR
通过量化 Fisher 信息矩阵确定的参数空间的几何,我们揭示了在深度神经网络中,批标准化作为最后一层标准化方法,对减少 Fisher 信息矩阵导致的病态锐度具有明显的贡献。
Abstract
normalization
methods play an important role in enhancing the performance of
deep learning
while their theoretical understandings have been limited. To theoretically elucidate the effectiveness of
→