neural networks 神經(jīng)網(wǎng)絡
activation function 激活函數(shù)
hyperbolic tangent 雙曲正切函數(shù)（tanh）
bias units 偏置項
activation 激活值
forward propagation 前向傳播
feedforward neural network 前饋神經(jīng)網(wǎng)絡
反向傳播算法 Backpropagation Algorithm
（批量）梯度下降法 (batch) gradient descent
（整體）代價函數(shù) (overall) cost function
方差 squared-error
均方差 average sum-of-squares error
規(guī)則化項 regularization term
權重衰減 weight decay
偏置項 bias terms
貝葉斯規(guī)則化方法 Bayesian regularization method
高斯先驗概率 Gaussian prior
極大后驗估計 MAP
極大似然估計 maximum likelihood estimation
激活函數(shù) activation function
雙曲正切函數(shù) tanh function
非凸函數(shù) non-convex function
隱藏層單元 hidden (layer) units
對稱失效 symmetry breaking
學習速率 learning rate
前向傳導 forward pass
假設值 hypothesis
殘差 error term
加權平均值 weighted average
前饋傳導 feedforward pass
阿達馬乘積 Hadamard product
前向傳播 forward propagation
