Add TERMINOLOGY.md (#302)

* Add TERMINOLOGY.md

* Remove extra space

README.md

@@ -91,95 +91,6 @@ make pdf
 python -m spacy download en # 需已 pip install spacy
 ```
 
-
-## 英汉术语对照
-
-
-accuracy，准确率
-
-baseline，基准
-
-batch，批量
-
-bias，偏差
-
-binary classification，二元分类
-
-bucketing，分桶
-
-class，类
-
-classification，分类
-
-collaborative filtering，协同过滤
-
-cost，成本
-
-cross-entropy，交叉熵
-
-data set，数据集
-
-decision boundary，决策边界
-
-dense，稠密
-
-dense layer，全连接层
-
-dropout，丢弃法
-
-empirical risk minimization，经验风险最小化
-
-epoch，迭代周期（周期）
-
-example，样本
-
-feature, 特征
-
-fully connected layer，全连接层
-
-hidden layer，隐藏层
-
-hidden variable，隐藏变量
-
-generalization，泛化
-
-hyperparameter，超参数
-
-hypothesis，假设
-
-import，导入
-
-independent and identically distributed(i.i.d)，独立同分布
-
-instance，实例
-
-label，标签
-
-logistic regression，逻辑回归
-
-mean squared error，均方误差
-
-metric，指标
-
-mini-batch，小批量
-
-normalization，归一化
-
-operator，运算符
-
-optimizer，优化器
-
-parameter，参数
-
-perplexity，困惑度
-
-pipeline，流水线
-
-size，大小
-
-transformation，变换
-
-
 ## 样式规范
 
 贡献请遵照本教程的[样式规范](FORMAT.md)。

TERMINOLOGY.md

+## 英汉术语对照
+
+鞍点, saddle point
+
+背景向量, context vector
+
+变换, transformation
+
+编码器，encoder
+
+标签, label
+
+步幅, stride
+
+参数, parameter
+
+超参数, hyper-parameter
+
+层序softmax, hierarchical softmax
+
+成本, cost
+
+词向量, embedding vector
+
+池化层, pooling layer
+
+稠密, dense
+
+大小, size
+
+导入, import
+
+迭代周期（周期), epoch
+
+迭代模型参数, update model parameter
+
+丢弃法, dropout
+
+动量法, momentum method
+
+独立同分布, independent and identically distributed(i.i.d.)
+
+端到端, end-to-end
+
+二元分类, binary classification
+
+发散, divergence
+
+泛化, generalization
+
+泛化误差, generalization error
+
+方差, variance
+
+非凸优化, non-convex optimization
+
+分桶, bucketing
+
+分类, classification
+
+分类器, classifier
+
+分词方式, tokenizer/tokenization
+
+负采样, negative sampling
+
+共现词频, n-gram frequency
+
+归一化, normalization
+
+过拟合, underfitting
+
+假设, hypothesis
+
+基准, baseline
+
+激活函数, activation function
+
+解码器, decoder
+
+经验风险最小化, empirical risk minimization
+
+局部最小值, local minimum
+
+卷积核, filter/kernel
+
+卷积神经网络, convolutional neural network
+
+决策边界, decision boundary
+
+均值, mean
+
+均方误差, mean squared error 
+
+困惑度, perplexity
+
+类, class
+
+交叉熵, cross-entropy
+
+连续词袋模型, Continous Bag of Words Model (CBOW)
+
+零张量, zero vector/tensor
+
+流水线, pipeline
+
+逻辑回归, logistic regression
+
+模型参数, model parameter
+
+模型复杂度, model complexity
+
+偏差, bias
+
+偏导数, partial derivative
+
+批量, batch
+
+平均池化层, mean pooling layer
+
+欠拟合, overfitting
+
+情感分析, sentiment analysis
+
+全连接层, fully connected layer/dense layer
+
+实例, instance
+
+随机均匀采样, uniform sampling
+
+收敛速度, convergence rate
+
+数值方法, numerical method
+
+数据集, data set
+
+损失函数, loss function
+
+双向循环神经网络, bidirectional recurrent neural network
+
+特征, feature
+
+特征值, eigenvalue
+
+梯度, gradient
+
+梯度裁剪, gradient clipping
+
+填充, padding
+
+跳字模型, skip-gram model
+
+调参, tune parameter
+
+通道, channel
+
+凸优化, convex optimization
+
+未知词符号, unknown token
+
+无偏估计, unbiased estimate
+
+小批量, mini-batch
+
+小批量梯度, mini-batch gradient
+
+线性模型, linear model
+
+线性回归模型, linear regression model
+
+协同过滤, collaborative filtering
+
+学习率, learning rate
+
+训练误差, optimization error
+
+循环神经网络, recurrent neural network
+
+样本, example
+
+一维梯度下降, gradient descent in one dimensional space
+
+隐藏变量, hidden variable
+
+隐藏层, hidden layer
+
+优化器, optimizer
+
+运算符, operator
+
+真实值, ground truth
+
+指标, metric
+
+支持向量机, support vector machine
+
+注意力机制模型, attention model
+
+准确率, accuracy
+
+最小化模型的损失函数, minimize loss function