人工智能词汇12.17

常用英语词汇-andrewNg课程intensity强度Regression回归Lossfunction损失函数non-convex非凸函数neuralnetwork神经网络supervisedlearning监督学习regressionproblem回归问题处理的是连续的问题classificationproblem分类问题discreetvalue离散值supportvectormachines支持向量机learningtheory学习理论learningalgorithms学习算法unsupervisedlearning无监督学习gradientdescent梯度下降linearregression线性回归NeuralNetwork神经网络gradientdescent梯度下降normalequationslinearalgebra线性代数superscript上标exponentiation指数trainingset训练集合trainingexample训练样本hypothesis假设，用来表示学习算法的输出LMSalgorithm“leastmeansquares最小二乘法算法batchgradientdescent批量梯度下降constantlygradientdescent随机梯度下降iterativealgorithm迭代算法partialderivative偏导数contour等高线quadraticfunction二元函数locallyweightedregression局部加权回归underfitting欠拟合overfitting过拟合non-parametriclearningalgorithms无参数学习算法parametriclearningalgorithm参数学习算法activation激活值activationfunction激活函数additivenoise加性噪声autoencoder自编码器Autoencoders自编码算法averagefiringrate平均激活率averagesum-of-squareserror均方差backpropagation后向传播basis基basisfeaturevectors特征基向量batchgradientascent批量梯度上升法Bayesianregularizationmethod贝叶斯规则化方法Bernoullirandomvariable伯努利随机变量biasterm偏置项binaryclassfication二元分类classlabels类型标记concatenation级联conjugategradient共轭梯度contiguousgroups联通区域convexoptimizationsoftware凸优化软件convolution卷积costfunction代价函数covariancematrix协方差矩阵DCcomponent直流分量decorrelation去相关degeneracy退化demensionalityreduction降维derivative导函数diagonal对角线diffusionofgradients梯度的弥散eigenvalue特征值eigenvector特征向量errorterm残差featurematrix特征矩阵featurestandardization特征标准化feedforwardarchitectures前馈结构算法feedforwardneuralnetwork前馈神经网络feedforwardpass前馈传导fine-tuned微调first-orderfeature一阶特征forwardpass前向传导forwardpropagation前向传播Gaussianprior高斯先验概率generativemodel生成模型gradientdescent梯度下降Greedylayer-wisetraining逐层贪婪训练方法groupingmatrix分组矩阵Hadamardproduct阿达马乘积HessianmatrixHessian矩阵hiddenlayer隐含层hiddenunits隐藏神经元Hierarchicalgrouping层次型分组higher-orderfeatures更高阶特征highlynon-convexoptimizationproblem高度非凸的优化问题histogram直方图hyperbolictangent双曲正切函数hypothesis估值，假设identityactivationfunction恒等激励函数IID独立同分布illumination照明inactive抑制independentcomponentanalysis独立成份分析inputdomains输入域inputlayer输入层intensity亮度/灰度interceptterm截距KLdivergence相对熵KLdivergenceKL分散度k-MeansK-均值learningrate学习速率leastsquares最小二乘法linearcorrespondence线性响应linearsuperposition线性叠加line-searchalgorithm线搜索算法localmeansubtraction局部均值消减localoptima局部最优解logisticregression逻辑回归lossfunction损失函数low-passfiltering低通滤波magnitude幅值MAP极大后验估计maximumlikelihoodestimation极大似然估计mean平均值MFCCMel倒频系数multi-classclassification多元分类neuralnetworks神经网络neuron神经元Newton’smethod牛顿法non-convexfunction非凸函数non-linearfeature非线性特征norm范式normbounded有界范数normconstrained范数约束normalization归一化numericalroundofferrors数值舍入误差numericallychecking数值检验numericallyreliable数值计算上稳定objectdetection物体检测objectivefunction目标函数off-by-oneerror缺位错误orthogonalization正交化outputlayer输出层overallcostfunction总体代价函数over-completebasis超完备基over-fitting过拟合partsofobjects目标的部件part-wholedecompostion部分-整体分解PCA主元分析penaltyterm惩罚因子per-examplemeansubtraction逐样本均值消减pooling池化pretrain预训练principalcomponentsanalysis主成份分析quadraticconstraints二次约束RBMs受限Boltzman机reconstructionbasedmodels基于重构的模型reconstructioncost重建代价reconstructionterm重构项redundant冗余reflectionmatrix反射矩阵regularization正则化regularizationterm正则化项rescaling缩放robust鲁棒性run行程second-orderfeature二阶特征sigmoidactivationfunctionS型激励函数significantdigits有效数字singularvalue奇异值singularvector奇异向量smoothedL1penalty平滑的L1范数惩罚SmoothedtopographicL1sparsitypenalty平滑地形L1稀疏惩罚函数smoothing平滑SoftmaxRegressonSoftmax回归sortedindecreasingorder降序排列sourcefeatures源特征sparseautoencoder消减归一化Sparsity稀疏性sparsityparameter稀疏性参数sparsitypenalty稀疏惩罚squarefunction平方函数squared-error方差stationary平稳性（不变性）stationarystochasticprocess平稳随机过程step-size步长值supervisedlearning监督学习symmetricpositivesemi-definitematrix对称半正定矩阵symmetrybreaking对称失效tanhfunction双曲正切函数theaverageactivation平均活跃度thederivativecheckingmethod梯度验证方法theempiricaldistribution经验分布函数theenergyfunction能量函数theLagrangedual拉格朗日对偶函数theloglikelihood对数似然函数thepixelintensityvalue像素灰度值therateofconvergence收敛速度topographiccostterm拓扑代价项topographicordered拓扑秩序transformation变换translationinvariant平移不变性trivialanswer平凡解under-completebasis不完备基unrolling组合扩展unsupervisedlearning无监督学习variance方差vecotrizedimplementation向量化实现vectorization矢量化visualcortex视觉皮层weightdecay权重衰减weightedaverage加权平均值whitening白化zero-mean均值为零Accumulatederrorbackpropagation累积误差逆传播ActivationFunction激活函数AdaptiveResonanceTheory/ART自适应谐振理论Addictivemodel加性学习AdversarialNetworks对抗网络AffineLayer仿射层Affinitymatrix亲和矩阵Agent代理/智能体Algorithm算法Alpha-betapruningα-β剪枝Anomalydetection异常检测Approximation近似AreaUnderROCCurve／AUCRoc曲线下面积ArtificialGeneralIntelligence/AGI通用人工智能ArtificialIntelligence/AI人工智能Associationanalysis关联分析Attentionmechanism注意力机制Attributeconditionalindependenceassumption属性条件独立性假设Attributespace属性空间Attributevalue属性值Autoencoder自编码器Automaticspeechrecognition自动语音识别Automaticsummarization自动摘要Averagegradient平均梯度Average-Pooling平均池化BackpropagationThroughTime通过时间的反向传播Backpropagation/BP反向传播Baselearner基学习器Baselearningalgorithm基学习算法BatchNormalization/BN批量归一化Bayesdecisionrule贝叶斯判定准则BayesModelAveraging／BMA贝叶斯模型平均Bayesoptimalclassifier贝叶斯最优分类器Bayesiandecisiontheory贝叶斯决策论