人工智能詞匯12.17

1、常用英語(yǔ)詞匯-andrewNg課程trainingset訓(xùn)練集合exponentiation指數(shù)averagefiringrate平均激活率intensity強(qiáng)度Regression回歸Lossfunction損失函數(shù)non-convex非凸函數(shù)neuralnetwork神經(jīng)網(wǎng)絡(luò)supervisedlearning監(jiān)督學(xué)習(xí)regressionproblem回歸問(wèn)題處理的是連續(xù)的問(wèn)題classificationproblem分類(lèi)問(wèn)題discreetvalue離散值supportvectormachines支持向量機(jī)learningtheory學(xué)習(xí)理論learningalgorithms學(xué)習(xí)算法u

2、nsupervisedlearning無(wú)監(jiān)督學(xué)習(xí)gradientdescent梯度下降linearregression線性回歸NeuralNetwork神經(jīng)網(wǎng)絡(luò)gradientdescent梯度下降normalequationslinearalgebra線性代數(shù)superscript上標(biāo)trainingexample訓(xùn)練樣本hypothesis假設(shè)，用來(lái)表示學(xué)習(xí)算法的輸出LMSalgorithm“l(fā)eastmeansquares最小二乘法算法batchgradientdescent批量梯度下降constantlygradientdescent隨機(jī)梯度下降iterativealgorithm迭代

3、算法partialderivative偏導(dǎo)數(shù)contour等高線quadraticfunction二元函數(shù)locallyweightedregression局部加權(quán)回歸underfitting欠擬合overfitting過(guò)擬合non-parametriclearningalgorithms無(wú)參數(shù)學(xué)習(xí)算法parametriclearningalgorithm參數(shù)學(xué)習(xí)算法activation激活值activationfunction激活函數(shù)additivenoise加性噪聲autoencoder自編碼器Autoencoders自編碼算法averagesum-of-squareserror均方差ba

4、ckpropagation后向傳播basis基basisfeaturevectors特征基向量batchgradientascent批量梯度上升法Bayesianregularizationmethod貝葉斯規(guī)則化方法Bernoullirandomvariable伯努利隨機(jī)變量biasterm偏置項(xiàng)binaryclassfication二元分類(lèi)classlabels類(lèi)型標(biāo)記concatenation級(jí)聯(lián)conjugategradient共軛梯度contiguousgroups聯(lián)通區(qū)域convexoptimizationsoftware凸優(yōu)化軟件convolution卷積costfunction

5、代價(jià)函數(shù)covariancematrix協(xié)方差矩陣DCcomponent直流分量decorrelation去相關(guān)degeneracy退化demensionalityreduction降維diagonal對(duì)角線diffusionofgradients梯度的彌散eigenvalue特征值eigenvector特征向量errorterm殘差featurematrix特征矩陣featurestandardization特征標(biāo)準(zhǔn)化feedforwardarchitectures前饋結(jié)構(gòu)算法feedforwardneuralnetwork前饋神經(jīng)網(wǎng)絡(luò)feedforwardpass前饋傳導(dǎo)fine-tune

6、d微調(diào)first-orderfeature一階特征forwardpass前向傳導(dǎo)forwardpropagation前向傳播Gaussianprior高斯先驗(yàn)概率generativemodel生成模型gradientdescent梯度下降Greedylayer-wisetraining逐層貪婪訓(xùn)練方法groupingmatrix分組矩陣Hadamardproduct阿達(dá)馬乘積HessianmatrixHessian矩陣hiddenlayer隱含層derivative導(dǎo)函數(shù)hiddenunits隱藏神經(jīng)元linearsuperposition線性疊加Hierarchicalgrouping層次型

7、分組higher-orderfeatures更高階特征highlynon-convexoptimizationproblem高度非凸的優(yōu)化問(wèn)題histogram直方圖hyperbolictangent雙曲正切函數(shù)hypothesis估值，假設(shè)identityactivationfunction恒等激勵(lì)函數(shù)IID獨(dú)立同分布illumination照明inactive抑制independentcomponentanalysis獨(dú)立成份分析inputdomains輸入域inputlayer輸入層intensity亮度/灰度interceptterm截距KLdivergence相對(duì)熵KLdiverge

8、nceKL分散度k-MeansK-均值learningrate學(xué)習(xí)速率leastsquares最小二乘法line-searchalgorithm線搜索算法localmeansubtraction局部均值消減localoptima局部最優(yōu)解logisticregression邏輯回歸lossfunction損失函數(shù)low-passfiltering低通濾波magnitude幅值MAP極大后驗(yàn)估計(jì)maximumlikelihoodestimation極大似然估計(jì)mean平均值MFCCMel倒頻系數(shù)multi-classclassification多元分類(lèi)neuralnetworks神經(jīng)網(wǎng)絡(luò)neur

9、on神經(jīng)元Newtonsmethod牛頓法non-convexfunction非凸函數(shù)non-linearfeature非線性特征norm范式normbounded有界范數(shù)normconstrained范數(shù)約束normalization歸一化linearcorrespondence線性響應(yīng)reconstructioncost重建代價(jià)numericalroundofferrors數(shù)值舍入誤差reconstructionbasedmodels基于重構(gòu)的模型sparseautoencoder消減歸一化numericallychecking數(shù)值檢驗(yàn)numericallyreliable數(shù)值計(jì)算上穩(wěn)定o

10、bjectdetection物體檢測(cè)objectivefunction目標(biāo)函數(shù)off-by-oneerror缺位錯(cuò)誤orthogonalization正交化outputlayer輸出層overallcostfunction總體代價(jià)函數(shù)over-completebasis超完備基over-fitting過(guò)擬合partsofobjects目標(biāo)的部件part-wholedecompostion部分-整體分解PCA主元分析penaltyterm懲罰因子per-examplemeansubtraction逐樣本均值消減pooling池化pretrain預(yù)訓(xùn)練principalcomponentsanal

11、ysis主成份分析quadraticconstraints二次約束RBMs受限Boltzman機(jī)reconstructionterm重構(gòu)項(xiàng)redundant冗余reflectionmatrix反射矩陣regularization正則化regularizationterm正則化項(xiàng)rescaling縮放robust魯棒性run行程second-orderfeature二階特征sigmoidactivationfunctionS型激勵(lì)函數(shù)significantdigits有效數(shù)字singularvalue奇異值singularvector奇異向量smoothedL1penalty平滑的L1范數(shù)懲罰S

12、moothedtopographicL1sparsitypenalty平滑地形L1稀疏懲罰函數(shù)smoothing平滑SoftmaxRegressonSoftmax回歸sortedindecreasingorder降序排列sourcefeatures源特征Sparsity稀疏性transformation變換Sparsity稀疏性transformation變換sparsityparameter稀疏性參數(shù)sparsitypenalty稀疏懲罰squarefunction平方函數(shù)squared-error方差stationary平穩(wěn)性（不變性）stationarystochasticprocess

13、平穩(wěn)隨機(jī)過(guò)程step-size步長(zhǎng)值supervisedlearning監(jiān)督學(xué)習(xí)symmetricpositivesemi-definitematrix對(duì)稱半正定矩陣symmetrybreaking對(duì)稱失效tanhfunction雙曲正切函數(shù)theaverageactivation平均活躍度thederivativecheckingmethod梯度驗(yàn)證方法theempiricaldistribution經(jīng)驗(yàn)分布函數(shù)theenergyfunction能量函數(shù)theLagrangedual拉格朗日對(duì)偶函數(shù)theloglikelihood對(duì)數(shù)似然函數(shù)thepixelintensityvalue像素灰

14、度值therateofconvergence收斂速度topographiccostterm拓?fù)浯鷥r(jià)項(xiàng)translationinvariant平移不變性trivialanswer平凡解under-completebasis不完備基unrolling組合擴(kuò)展unsupervisedlearning無(wú)監(jiān)督學(xué)習(xí)variance方差vecotrizedimplementation向量化實(shí)現(xiàn)vectorization矢量化visualcortex視覺(jué)皮層weightdecay權(quán)重衰減weightedaverage加權(quán)平均值whitening白化zero-mean均值為零Accumulatederrorba

15、ckpropagation累積誤差逆?zhèn)鞑ctivationFunction激活函數(shù)AdaptiveResonanceTheory/ART自適應(yīng)諧振理論Addictivemodel加性學(xué)習(xí)AdversarialNetworks對(duì)抗網(wǎng)絡(luò)AffineLayer仿射層Affinitymatrix親和矩陣Agent代理/智能體topographicordered拓?fù)渲刃駽ost-sensitive成本敏感Algorithm算法BatchNormalization/BN批量歸一化Alpha-betapruninga-p剪枝Anomalydetection異常檢測(cè)Approximation近似AreaUn

16、derROCCurveAUCRoc曲線下面積ArtificialGeneralIntelligence/AGI通用人工智能ArtificialIntelligence/AI人工智能Associationanalysis關(guān)聯(lián)分析Attentionmechanism注意力機(jī)制Attributeconditionalindependenceassumption屬性條件獨(dú)立性假設(shè)Attributespace屬性空間Attributevalue屬性值A(chǔ)utoencoder自編碼器Automaticspeechrecognition自動(dòng)語(yǔ)音識(shí)別Automaticsummarization自動(dòng)摘要Avera

17、gegradient平均梯度Average-Pooling平均池化BackpropagationThroughTime通過(guò)時(shí)間的反向傳播Backpropagation/BP反向傳播Baselearner基學(xué)習(xí)器Bayesdecisionrule貝葉斯判定準(zhǔn)則BayesModelAveragingBMA貝葉斯模型平均Bayesoptimalclassifier貝葉斯最優(yōu)分類(lèi)器Bayesiandecisiontheory貝葉斯決策論Bayesiannetwork貝葉斯網(wǎng)絡(luò)Between-classscattermatrix類(lèi)間散度矩陣Bias偏置/偏差Bias-variancedecomposit

18、ion偏差-方差分解Bias-VarianeeDilemma偏差一方差困境Bi-directionalLong-ShortTermMemory/Bi-LSTM雙向長(zhǎng)短期記憶Binaryelassifieation二分類(lèi)Binomialtest二項(xiàng)檢驗(yàn)Bi-partition二分法Boltzmannmaehine玻爾茲曼機(jī)Bootstrapsampling自助采樣法可重復(fù)采樣Bootstrapping自助法Break-EventPointBEP平衡點(diǎn)Calibration校準(zhǔn)Caseade-Correlation級(jí)聯(lián)相關(guān)Categoriealattribute離散屬性Baselearningal

19、gorithm基學(xué)習(xí)算法Class-eonditionalprobability類(lèi)條件概率Classificationandregressiontree/CART分類(lèi)與回ConditionalProbabilityTableCPT條件概率表歸樹(shù)Classifier分類(lèi)器Class-imbalance類(lèi)別不平衡Closed-form閉式Cluster簇/類(lèi)/集群Clusteranalysis聚類(lèi)分析Clustering聚類(lèi)Clusteringensemble聚類(lèi)集成Co-adapting共適應(yīng)Codingmatrix編碼矩陣COLT國(guó)際學(xué)習(xí)理論會(huì)議Committee-basedlearning基

20、于委員會(huì)的學(xué)習(xí)Competitivelearning競(jìng)爭(zhēng)型學(xué)習(xí)Componentlearner組件學(xué)習(xí)器Comprehensibility可解釋性ComputationCost計(jì)算成本ComputationalLinguistics計(jì)算語(yǔ)言學(xué)Computervision計(jì)算機(jī)視覺(jué)Conceptdrift概念漂移ConceptLearningSystem/CLS概念學(xué)習(xí)系統(tǒng)Conditionalentropy條件熵Conditionalrandomfield/CRF條件隨機(jī)場(chǎng)Conditionalrisk條件風(fēng)險(xiǎn)Confidence置信度Confusionmatrix混淆矩陣Connection

21、weight連接權(quán)Connectionism連結(jié)主義Consistency一致性相合性Contingencytable列聯(lián)表Continuousattribute連續(xù)屬性Convergence收斂Conversationalagent會(huì)話智能體Convexquadraticprogramming凸二次規(guī)劃Convexity凸性Convolutionalneuralnetwork/CNN卷積神經(jīng)網(wǎng)絡(luò)Co-occurrence同現(xiàn)Correlationcoefficient相關(guān)系數(shù)Cosinesimilarity余弦相似度Costcurve成本曲線CostFunction成本函數(shù)Costmatri

22、x成本矩陣Conditionalmutualinformation條件互信息Crossentropy交叉熵Crossvalidation交叉驗(yàn)證Crowdsourcing眾包Curseofdimensionality維數(shù)災(zāi)難Cutpoint截?cái)帱c(diǎn)Cuttingplanealgorithm割平面法Datamining數(shù)據(jù)挖掘Dataset數(shù)據(jù)集DecisionBoundary決策邊界Decisionstump決策樹(shù)樁Decisiontree決策樹(shù)判定樹(shù)Deduction演繹DeepBeliefNetwork深度信念網(wǎng)絡(luò)DeepConvolutionalGenerativeAdversarialN

23、etworkDCGAN深度卷積生成對(duì)抗網(wǎng)絡(luò)Deeplearning深度學(xué)習(xí)Deepneuralnetwork/DNN深度神經(jīng)網(wǎng)絡(luò)DeepQ-Learning深度Q學(xué)習(xí)DeepQ-Network深度Q網(wǎng)絡(luò)Densityestimation密度估計(jì)Density-basedclustering密度聚類(lèi)Differentiableneuralcomputer可微分神經(jīng)計(jì)算機(jī)Dimensionalityreductionalgorithm降維算法Directededge有向邊Disagreementmeasure不合度量Discriminativemodel判別模型Discriminator判別器Di

24、stancemeasure距離度量Distancemetriclearning距離度量學(xué)習(xí)Distribution分布Divergence散度Diversitymeasure多樣性度量差異性度量Domainadaption領(lǐng)域自適應(yīng)Downsampling下采樣D-separation(Directedseparation)有向分離Dualproblem對(duì)偶問(wèn)題Dummynode啞結(jié)點(diǎn)DynamicFusion動(dòng)態(tài)融合Dynamicprogramming動(dòng)態(tài)規(guī)劃Eigenvaluedecomposition特征值分解Embedding嵌入Emotionalanalysis情緒分析Empiric

25、alconditionalentropy經(jīng)驗(yàn)條件熵FalsePositiveRate/FPR假正例率FalsePositiveRate/FPR假正例率Empiricalentropy經(jīng)驗(yàn)熵Featureselection特征選擇Empiricalerror經(jīng)驗(yàn)誤差Empiricalrisk經(jīng)驗(yàn)風(fēng)險(xiǎn)End-to-End端到端Energy-basedmodel基于能量的模型Ensemblelearning集成學(xué)習(xí)Ensemblepruning集成修剪ErrorCorrectingOutputCodesECOC糾錯(cuò)輸出碼Errorrate錯(cuò)誤率Error-ambiguitydecomposition

26、誤差-分歧分解Euclideandistance歐氏距離Evolutionarycomputation演化計(jì)算Expectation-Maximization期望最大化Expectedloss期望損失ExplodingGradientProblem梯度爆炸問(wèn)題Exponentiallossfunction指數(shù)損失函數(shù)ExtremeLearningMachine/ELM超限學(xué)習(xí)機(jī)Factorization因子分解Falsenegative假負(fù)類(lèi)Falsepositive假正類(lèi)Featurevector特征向量FeaturedLearning特征學(xué)習(xí)FeedforwardNeuralNetwork

27、s/FNN前饋神經(jīng)網(wǎng)絡(luò)Fine-tuning微調(diào)Flippingoutput翻轉(zhuǎn)法Fluctuation震蕩Forwardstagewisealgorithm前向分步算法Frequentist頻率主義學(xué)派Full-rankmatrix滿秩矩陣Functionalneuron功能神經(jīng)元Gainratio增益率Gametheory博弈論Gaussiankernelfunction高斯核函數(shù)GaussianMixtureModel高斯混合模型GeneralProblemSolving通用問(wèn)題求解Generalization泛化Generalizationerror泛化誤差Generalization

28、errorbound泛化誤差上界GeneralizedLagrangefunction廣義拉格朗日函數(shù)Generalizedlinearmodel廣義線性模型GeneralizedRayleighquotient廣義瑞利商Featureengineering特征工程GenerativeAdversarialNetworks/GAN生成對(duì)抗Hold-out留出法GenerativeAdversarialNetworks/GAN生成對(duì)抗Hold-out留出法Hingelossfunction合頁(yè)損失函數(shù)Insensitiveloss不敏感損失Hingelossfunction合頁(yè)損失函數(shù)Insen

29、sitiveloss不敏感損失網(wǎng)絡(luò)Homogeneous同質(zhì)Hybridcomputing混合計(jì)算Hyperparameter超參數(shù)Hypothesis假設(shè)Hypothesistest假設(shè)驗(yàn)證ICML國(guó)際機(jī)器學(xué)習(xí)會(huì)議Improvediterativescaling/IIS改進(jìn)的迭代尺度法Incrementallearning增量學(xué)習(xí)Independentandidenticallydistributed/i.i.d.獨(dú)立同分布IndependentComponentAnalysis/ICA獨(dú)立成分分析Indicatorfunction指示函數(shù)Individuallearner個(gè)體學(xué)習(xí)器Indu

30、ction歸納Inductivebias歸納偏好Inductivelearning歸納學(xué)習(xí)InductiveLogicProgrammingILP歸納邏輯程序設(shè)計(jì)Informationentropy信息熵Informationgain信息增益Inputlayer輸入層GenerativeModel生成模型Generator生成器GeneticAlgorithm/GA遺傳算法Gibbssampling吉布斯采樣Giniindex基尼指數(shù)Globalminimum全局最小GlobalOptimization全局優(yōu)化Gradientboosting梯度提升GradientDescent梯度下降Graphtheory圖論Ground-truth真相真實(shí)Hardmargin硬間隔Hardvoting硬投票Harmonicmean調(diào)和平均Hessematrix海塞矩陣Hiddendynamicmodel隱動(dòng)態(tài)模型Hiddenlayer隱藏層HiddenMarkovModel/HMM隱馬爾可夫模型Hierarchicalclustering層次聚類(lèi)Hilbertspace希爾伯特空間Inter-