统计学一些名称中英文对照
奥卡姆剃刀:Occam’s razor
半监督学习:semi-supervised learning
标注:tagging
不完全数据:incomplete-data
参数空间:parameter space
残差:residual
测试集:test set
测试数据:test data
测试误差:test error
策略:strategy
成对马可尔可夫性:pairwise Markov property
词性标注:part of speech tagging
代价函数:cost function
代理损失函数:surrogate loss function
带符号的距离:signed distance
动态规划:dynamjc programming
对偶算法:dual algorithm
对偶问题:dual problem
对数几率:log odds
对数似然损失函数:log-likelihood loss function
对数损失函数:logarithmic loss function
对数线性模型:log linear model
多数表决规则:majority voting rule
多项逻辑斯蒂回归模型:multi-nominal logistic regression model
多项式核函数:polynominal kenel function
二项逻辑斯蒂回归模型:binomial logistic regression model
罚项:penalty term
泛化能力:generalization ability
泛化误差:generalization error
泛化误差上界:generalization error bound
非监督学习:unsupervised learning
非线性支持向量机:non-linear support vector machine
分类:clssification
分类器:classifier
分类与回归树:clssification and regression tree, CART
分离超平面:speparating hyperplane
风险函数:risk function
改进的迭代尺度法:inproved iterative scaling ,IIS
概率近似正确:probabilistic approximately correct , PAC
概率无向图模型:probabilistic undirected graphical model
感知机:perceptron
高斯核函数:Gaussian kernel function
高斯混合模型:Gaussian mixture model
根节点:root node
估计误差:estimation error
观测变量:observable variable
观测序列:observation sequence
广义拉格朗日函数:generalized Lagrange function
广义期望极大:generalized expectation maximization ,GEM
规范化因子:normalization factor
过拟合:over-fitting
海赛矩阵:Hesse matrix
函数间隔:function margin
合页损失函数:hinge loss function
核方法:kernel method
核函数:kernel function
互信息:mutual information
划分:partition
回归:regresion
基尼系数:Gini index
极大-极大算法:maximization-maximization algorithm
极大似然估计:maximum likelihood estimation
几何间隔:geometric margin
几率:odds
加法模型:additive model
假设空间:hypothesis space
间隔:margin
监督学习:supervised learning
剪枝:pruning
交叉验证:cross validation
节点:node
结构风险最小化:structural risk minimization ,SRM
解码:decoding
近似误差:approximation error
经验风险:empirical entropy
经验风险最小化:empirical risk minimization , ERM
经验熵:empirical emtropy
经验损失:empirical loss
经验条件熵:empirical conditional entropy
精确率:precision
径向基函数:radial basis function
局部马尔科夫性:local Markov property
决策函数:decision function
决策树:decision tree
决策树桩:decision stump
绝对损失函数:absolute loss function
拉格朗日乘子:Lagrange multiplier
拉格朗日对偶性:Lagrange duality
拉格朗日函数:Lagrange function
拉普拉斯平滑:Laplace smoothing
类:class
类标记:class label
留一交叉验证:leave-one-out cross validation
逻辑斯蒂分布:logistic distribution
逻辑斯蒂回归:logistic regression
马尔科夫随机场:Markov random field
曼哈顿距离:Manhattan distance
模型:model
模型选择:model selection
内部节点:internal model
纳特:nat
拟牛顿法:quasi Newton method
牛顿法:Newton method
欧氏距离:Euclidean distance
判别方法:discrimination approach
判别模型:discriminative model
偏置:bias
平方损失函数:quadratic loss function
评价准则:evaluation criterion
朴素贝叶斯:na?ve Bayes
朴素贝叶斯算法:na?ve Bayes algorithm
期望极大算法(EM算法):expectations maximization algorithm
期望损失:expected loss
前向分布算法:forward stagewise algorithm
前向-后向算法:forward-backward algorithm
潜在变量:latent variable
强化学习:reinforcement learning
强可学习:strongly learnable
切分变量:splitting variable
切分点:splitting point
全局马尔科夫性:global Markov property
权值:weight
权值向量:weight vector
软间隔最大化:soft margin maximization
弱可学习:weakly learnable
熵:entropy
生成方法:generative approach
生成模型:generative model
实例:instance
势函数:potential function
输出空间:output space
输入空间:input space
数据:data
算法:algorithm
随机梯度下降法:stochastic gradient descent
损失函数:loss function
特异点:outlier
特征函数:feature function
特征空间:feature space
特征向量:feature vector
梯度提升:gradient boosting
梯度下降法:gradient descent
提升:boosting
提升树:boosting tree
提早停止:early stopping
条件熵:conditional entropy
条件随机场:conditional randomfield , CRF
统计机器学习:statistical machine learning
统计学习:statistical learning
统计学习方法:statistical learning method
统计学习理论:statistical learning theory
统计学习应用:application of statistical learning
凸二次规划:convex quadratic programming
图:graph
团:clique
完全数据:complete-data
维特比算法:Viterbi algorithm
文本分类:text classification
误差率:error rate
希尔伯特空间:Hilbert space
线性分类模型:linear classification model
线性分类器:linear classifier
线性可分数据集:linearly separable data set
线性可分支持向量机:linear support vector machine in linearly separable case
线性链:linear chain
线性链条件随机场:linear chain conditional random field
线性扫描:linear scan
线性支持向量机:linear support vector machine
信息增益:information gain
信息增益比:information gain ratio
序列最小最优化:sequential minimal optimization
学习率:learning rate
训练集:training set
训练数据:training data
训练误差:training error
验证集:validation set
叶节点:leaf node
因子分解:factorization
隐变量:hidden variable
隐马尔可夫模型:hidden Markov model, HMM
硬间隔最大化:hard margin maximization
有向边:directed edge
余弦相似度:cosine similarity
预测:prediction
原始问题:primal problem
再生核希尔伯特空间:reproducing kernel Hilbert space,RKHS
召回率:recall
正定核函数:positive definite kernel function
正则化:regularization
正则化项:regularizer
支持向量:support vector
支持向量机:support vector machine ,SVM
指示函数:indicator function
指示损失函数:exponential loss function
中位数:median
状态序列:state sequence
准确率:accuracy
字符串核函数:string kernel function
最大后验概率估计:maximum posterior probability estimation, MAP
最大间隔法:maximum margin method
最大熵模型:maximum entropy method
最大团:maximal clique
最速下降法:steepest descent
最小二乘法:least squares
最小二乘回归树:least squares regression tree
0-1损失函数:0-1 loss function
Adaboost算法:AdaBoost algorithm
Baum-Welch算法:
BFGS算法:Broyden-Fletcher-Goldfarb-Shanno algorithm ,BFGS algorithm
Broyden类算法:Broyden’s algorithm
C4.5算法:
DFP算法:Davidon-Fletcher-Powell algorithm
EM算法、
F函数、
Gram矩阵:Gram matrix
ID3算法、
Jensen不等式:Jensen inequality
Kd树、
K近邻法:K-nearest neighbor,K-NN
L1范数:L1 norm
L2范数:L2 norm
Lp范数:Lp distance
Minkowski距离:Minkowski distance
Q函数、
S形曲线:sigmoid curve
S折交叉验证:S-fold cross validation