用反向传导思想求导
From Ufldl
(→中文译者) |
|||
Line 1: | Line 1: | ||
- | |||
- | |||
== 简介 == | == 简介 == | ||
Line 312: | Line 310: | ||
\end{align} | \end{align} | ||
</math> | </math> | ||
+ | |||
+ | |||
+ | |||
+ | ==中英文对照== | ||
+ | |||
+ | :反向传导 backpropagation | ||
+ | :稀疏编码 sparse coding | ||
+ | :权重矩阵 weight matrix | ||
+ | :目标函数 objective | ||
+ | :平滑地形L1稀疏罚函数 Smoothed topographic L1 sparsity penalty | ||
+ | :重建代价 reconstruction cost | ||
+ | :稀疏自编码器 sparse autoencoder | ||
+ | :梯度 gradient | ||
+ | :神经网络 neural network | ||
+ | :神经元 neuron | ||
+ | :激励 activation | ||
+ | :激励函数 activation function | ||
+ | :独立成分分析 independent component analysis | ||
+ | :单位激励函数 identity activation function | ||
+ | :平方函数 square function | ||
+ | :分组矩阵 grouping matrix | ||
+ | :特征矩阵 feature matrix | ||
Line 317: | Line 337: | ||
==中文译者== | ==中文译者== | ||
- | @ | + | 葛燕儒(yrgehi@gmail.com), 顾祺龙(ggnle@hotmail.com), 李良玥(jackiey99@gmail.com), 王方(fangkey@gmail.com) |
+ | |||
+ | |||
+ | |||
+ | {{Languages|Deriving_gradients_using_the_backpropagation_idea|English}} |