累積誤差逆?zhèn)鞑ダ鄯e誤差逆?zhèn)鞑ゼせ詈瘮?shù)自適應諧振理論加性學習對抗網(wǎng)絡仿射層親和矩陣代理/智能體算法-剪枝異常檢測近似Roc曲線下面積通用人工智能人工智能關(guān)聯(lián)分析注意力機制屬性條件獨立性假設屬性空間屬性值自編碼器自動語音識別自動摘要平均梯度平均池化LetterAAccumulatederrorbackpropagationActivationFunctionAdaptiveResonanceTheory/ARTAddictivemodelAdversarialNetworksAffineLayerAffinitymatrixAgentAlgorithmAlpha-betapruningAnomalydetectionApproximationAreaUnderROCCurve／AUCArtificialGeneralIntelligence/AGIArtificialIntelligence/AIAssociationanalysisAttentionmechanismAttributeconditionalindependenceassumptionAttributespaceAttributevalueAutoencoderAutomaticspeechrecognitionAutomaticsummarizationAveragegradientAverage-PoolingLetterBBackpropagationThroughTimeBackpropagation/BPBaselearnerBaselearningalgorithmBatchNormalization/BNBayesdecisionruleBayesModelAveraging／BMABayesoptimalclassifierBayesiandecisiontheoryBayesiannetworkBetween-classscattermatrixBiasBias-variancedecompositionBias-VarianceDilemmaBi-directionalLong-ShortTermMemory/Bi-LSTMBinaryclassificationBinomialtestBi-partitionBoltzmannmachineBootstrapsamplingBootstrappingBreak-EventPoint／BEPLetterCCalibrationCascade-CorrelationCategoricalattribute通過時間的反向傳播反向傳播基學習器基學習算法批量歸一化貝葉斯判定準則貝葉斯模型平均貝葉斯最優(yōu)分類器貝葉斯決策論貝葉斯網(wǎng)絡類間散度矩陣偏置/偏差偏差-方差分解偏差方差困境雙向長短期記憶二分類二項檢驗二分法玻爾茲曼機自助采樣法／可重復采樣／有放回采樣自助法平衡點校準級聯(lián)相關(guān)離散屬性

Class-conditionalprobabilityClassificationandregressiontree/CARTClassifierClass-imbalanceClosed-formClusterClusteranalysisClusteringClusteringensembleCo-adaptingCodingmatrixCOLTCommittee-basedlearningCompetitivelearningComponentlearnerComprehensibilityComputationCostComputationalLinguisticsComputervisionConceptdriftConceptLearningSystem/CLSConditionalentropyConditionalmutualinformation類條件概率分類與回歸樹分類器類別不平衡閉式簇/類/集群聚類分析聚類ConditionalProbabilityTable／條件概率表CPTConditionalProbabilityTable／條件概率表CPTConditionalrandomfield/CRF條件隨機場Conditionalrisk 條件風險ConfidenceConfusionmatrixConnectionweightConnectionismConsistencyContingencytableContinuousattributeConvergenceConversationalagentConvexquadraticprogrammingConvexityConvolutionalneuralnetwork/CNNCo-occurrenceCorrelationcoefficientCosinesimilarityCostcurveCostFunctionCostmatrixCost-sensitiveCrossentropyCrossvalidationCrowdsourcingCurseofdimensionalityCutpointCuttingplanealgorithmLetterDDatamining置信度混淆矩陣連接權(quán)連結(jié)主義一致性／相合性列聯(lián)表連續(xù)屬性收斂會話智能體凸二次規(guī)劃凸性卷積神經(jīng)網(wǎng)絡同現(xiàn)相關(guān)系數(shù)余弦相似度成本曲線成本函數(shù)成本矩陣成本敏感交叉熵交叉驗證眾包維數(shù)災難截斷點割平面法數(shù)據(jù)挖掘DatasetDecisionBoundaryDecisionstumpDecisiontreeDeductionDeepBeliefNetworkDeepConvolutionalGenerativeAdversarialNetwork/DCGANDeeplearningDeepneuralnetwork/DNNDeepQ-LearningDeepQ-NetworkDensityestimationDensity-basedclusteringDifferentiableneuralcomputerDimensionalityreductionalgorithmDirectededgeDisagreementmeasureDiscriminativemodelDiscriminatorDistancemeasureDistancemetriclearningDistributionDivergenceDiversitymeasureDomainadaptionDownsampling數(shù)據(jù)集決策邊界決策樹樁決策樹／判定樹演繹深度信念網(wǎng)絡深度卷積生成對抗網(wǎng)絡深度學習深度神經(jīng)網(wǎng)絡深度Q學習深度Q網(wǎng)絡密度估計密度聚類可微分神經(jīng)計算機降維算法有向邊不合度量判別模型判別器距離度量距離度量學習分布散度多樣性度量／差異性度量領域自適應下采樣D-separation(Directedseparation)DualproblemDummynodeDynamicFusionDynamicprogrammingLetterEEigenvaluedecompositionEmbeddingEmotionalanalysisEmpiricalconditionalentropyEmpiricalentropyEmpiricalerrorEmpiricalriskEnd-to-EndEnergy-basedmodelEnsemblelearningEnsemblepruningErrorCorrectingOutputCodes／ECOCErrorrateError-ambiguitydecompositionEuclideandistanceEvolutionarycomputationExpectation-MaximizationExpectedlossExplodingGradientProblemExponentiallossfunction有向分離對偶問題啞結(jié)點動態(tài)融合動態(tài)規(guī)劃特征值分解嵌入情緒分析經(jīng)驗條件熵經(jīng)驗熵經(jīng)驗誤差經(jīng)驗風險端到端基于能量的模型集成學習集成修剪糾錯輸出碼錯誤率誤差-分歧分解歐氏距離演化計算期望最大化期望損失梯度爆炸問題指數(shù)損失函數(shù)超限學習機超限學習機因子分解假負類假正類假正例率特征工程特征選擇特征向量特征學習前饋神經(jīng)網(wǎng)絡微調(diào)翻轉(zhuǎn)法震蕩前向分步算法頻率主義學派滿秩矩陣功能神經(jīng)元增益率博弈論高斯核函數(shù)高斯混合模型通用問題求解泛化泛化誤差泛化誤差上界ExtremeLearningMachine/ELMLetterFFactorizationFalsenegativeFalsepositiveFalsePositiveRate/FPRFeatureengineeringFeatureselectionFeaturevectorFeaturedLearningFeedforwardNeuralNetworks/FNNFine-tuningFlippingoutputFluctuationForwardstagewisealgorithmFrequentistFull-rankmatrixFunctionalneuronLetterGGainratioGametheoryGaussiankernelfunctionGaussianMixtureModelGeneralProblemSolvingGeneralizationGeneralizationerrorGeneralizationerrorboundGeneralizedLagrangefunctionGeneralizedlinearmodelGeneralizedRayleighquotientGenerativeAdversarialNetworks/GANGenerativeModelGeneratorGeneticAlgorithm/GAGibbssamplingGiniindexGlobalminimumGlobalOptimizationGradientboostingGradientDescentGraphtheoryGround-truthLetterHHardmarginHardvotingHarmonicmeanHessematrixHiddendynamicmodelHiddenlayerHiddenMarkovModel/HMMHierarchicalclusteringHilbertspaceHingelossfunctionHold-out廣義拉格朗日函數(shù)廣義線性模型廣義瑞利商生成對抗網(wǎng)絡生成模型生成器遺傳算法吉布斯采樣基尼指數(shù)全局最小全局優(yōu)化梯度提升梯度下降圖論真相／真實硬間隔硬投票調(diào)和平均海塞矩陣隱動態(tài)模型隱藏層隱馬爾可夫模型層次聚類希爾伯特空間合頁損失函數(shù)留出法Homogeneous同質(zhì)Hybridcomputing混合計算Hyperparameter超參數(shù)Hypothesis假設HypothesistestLetterI假設驗證ICML國際機器學習會議Improvediterativescaling/IIS改進的迭代尺度法IncrementallearningIndependentandidentically增量學習獨立同分布distributed/i.i.d.IndependentComponentAnalysis/ICA獨立成分分析Indicatorfunction指示函數(shù)Individuallearner個體學習器Induction歸納Inductivebias歸納偏好Inductivelearning歸納學習InductiveLogicProgramming／ILP歸納邏輯程序設計Informationentropy信息熵Informationgain信息增益Inputlayer輸入層Insensitiveloss不敏感損失Inter-clustersimilarity簇間相似度InternationalConferenceforMachine國際機器學習大會Learning/ICMLIntra-clustersimilarity簇內(nèi)相似度Intrinsicvalue固有值IsometricMapping/IsomapIsotonicregressionIterativeDichotomiserLetterKKernelmethodKerneltrickKernelizedLinearDiscriminantAnalysis／KLDAK-foldcrossvalidationK-MeansClusteringK-NearestNeighboursAlgorithm/KNNKnowledgebaseKnowledgeRepresentationLetterLLabelspaceLagrangedualityLagrangemultiplierLaplacesmoothingLaplaciancorrectionLatentDirichletAllocationLatentsemanticanalysisLatentvariableLazylearningLearnerLearningbyanalogyLearningrate等度量映射等分回歸迭代二分器核方法核技巧核線性判別分析k折交叉驗證/k倍交叉驗證K 均值聚類K近鄰算法知識庫知識表征標記空間拉格朗日對偶性拉格朗日乘子拉普拉斯平滑拉普拉斯修正隱狄利克雷分布潛在語義分析隱變量懶惰學習學習器類比學習學習率LearningVectorQuantization/LVQ學習向量量化Leastsquaresregressiontree 最小二乘回歸樹Leave-One-Out/LOO 留一法linearchainconditionalrandomfield線性鏈條件隨機場LinearDiscriminantAnalysis／LDA線性判別分析Linearmodel線性模型LinearRegression線性回歸Linkfunction聯(lián)系函數(shù)LocalMarkovproperty局部馬爾可夫性Localminimum局部最小Loglikelihood對數(shù)似然Logodds／logit對數(shù)幾率LogisticRegressionLogistic回歸Log-likelihood對數(shù)似然Log-linearregression對數(shù)線性回歸Long-ShortTermMemory/LSTM長短期記憶Lossfunction損失函數(shù)LetterMMachinetranslation/MT機器翻譯Macron-P宏查準率Macron-R宏查全率Majorityvoting絕對多數(shù)投票法Manifoldassumption流形假設Manifoldlearning流形學習Margintheory間隔理論Marginaldistribution邊際分布Marginalindependence邊際獨立性Marginalization邊際化MarkovChainMonteCarlo/MCMC馬爾可夫鏈蒙特卡羅方法

MarkovRandomFieldMaximalcliqueMaximumLikelihoodEstimation/MLEMaximummarginMaximumweightedspanningtreeMax-PoolingMeansquarederrorMeta-learnerMetriclearningMicro-PMicro-RMinimalDescriptionLength/MDLMinimaxgameMisclassificationcostMixtureofexpertsMomentumMoralgraphMulti-classclassificationMulti-documentsummarizationMulti-layerfeedforwardneuralnetworksMultilayerPerceptron/MLPMultimodallearningMultipleDimensionalScalingMultiplelinearregressionMulti-responseLinearRegression／馬爾可夫隨機場最大團極大似然估計／極大似然法馬爾可夫隨機場最大團極大似然估計／極大似然法最大間隔最大帶權(quán)生成樹最大池化均方誤差元學習器度量學習微查準率微查全率最小描述長度極小極大博弈誤分類成本混合專家動量道德圖／端正圖多分類多文檔摘要多層前饋神經(jīng)網(wǎng)絡多層感知器多模態(tài)學習多維縮放多元線性回歸多響應線性回歸MutualinformationLetterNNaivebayesNaiveBayesClassifierNamedentityrecognitionNashequilibriumNaturallanguagegeneration/NLGNaturallanguageprocessingNegativeclassNegativecorrelationNegativeLogLikelihoodNeighbourhoodComponentAnalysis/NCANeuralMachineTranslationNeuralTuringMachineNewtonmethodNIPSNoFreeLunchTheorem／NFLNoise-contrastiveestimationNominalattributeNon-convexoptimizationNonlinearmodelNon-metricdistanceNon-negativematrixfactorizationNon-ordinalattributeNon-SaturatingGameNorm互信息樸素貝葉斯樸素貝葉斯分類器互信息樸素貝葉斯樸素貝葉斯分類器命名實體識別納什均衡自然語言生成自然語言處理負類負相關(guān)法負對數(shù)似然近鄰成分分析神經(jīng)機器翻譯神經(jīng)圖靈機牛頓法國際神經(jīng)信息處理系統(tǒng)會議沒有免費的午餐定理噪音對比估計列名屬性非凸優(yōu)化非線性模型非度量距離非負矩陣分解無序?qū)傩苑秋柡筒┺姆稊?shù)歸一化核范數(shù)數(shù)值屬性核范數(shù)數(shù)值屬性NumericalattributeLetterOObjectivefunction 目標函數(shù)Obliquedecisiontree 斜決策樹Occam’srazor 奧卡姆剃刀Odds 幾率Off-Policy 離策略Oneshotlearning 一次性學習One-DependentEstimator／ODE獨依賴估計On-Policy在策略Ordinalattribute有序?qū)傩設ut-of-bagestimate包外估計Outputlayer輸出層Outputsmearing輸出調(diào)制法Overfitting過擬合／過配Oversampling過采樣LetterPPairedt-test成對t檢驗Pairwise成對型PairwiseMarkovproperty成對馬爾可夫性Parameter參數(shù)Parameterestimation參數(shù)估計Parametertuning調(diào)參Parsetree解析樹ParticleSwarmOptimization/PSO粒子群優(yōu)化算法Part-of-speechtagging 詞性標注Perceptron 感知機Performancemeasure性能度量相對多數(shù)投票法極性檢測多項式核函數(shù)池化正類正定矩陣相對多數(shù)投票法極性檢測多項式核函數(shù)池化正類正定矩陣后續(xù)檢驗后剪枝勢函數(shù)查準率／準確率預剪枝先驗概率圖模型近端梯度下降剪枝偽標記量子化神經(jīng)網(wǎng)絡量子計算機量子計算擬牛頓法徑向基函數(shù)隨機森林算法PlugandPlayGenerativeNetwork即插即用生成網(wǎng)絡PluralityvotingPolaritydetectionPolynomialkernelfunctionPoolingPositiveclassPositivedefinitematrixPost-hoctestPost-pruningpotentialfunctionPrecisionPrepruningPrincipalcomponentanalysis/PCA主成分分析Principleofmultipleexplanations多釋原則PriorProbabilityGraphicalModelProximalGradientDescent/PGDPruningPseudo-labelLetterQQuantizedNeuralNetworkQuantumcomputerQuantumComputingQuasiNewtonmethodLetterRRadialBasisFunction／RBFRandomForestAlgorithmRestrictedIsometryProperty/RIPRestrictedIsometryProperty/RIPRandomwalkRecallReceiverOperatingCharacteristic/ROCRectifiedLinearUnit/ReLURecurrentNeuralNetworkRecursiveneuralnetworkReferencemodelRegressionRegularizationReinforcementlearning/RLRepresentationlearningRepresentertheoremreproducingkernelHilbertspace/RKHSRe-samplingRescalingResidualMappingResidualNetworkRestrictedBoltzmannMachine/RBMRe-weightingRobustnessRootnodeRuleEngineRulelearningLetterS隨機漫步查全率／召回率受試者工作特征線性修正單元循環(huán)神經(jīng)網(wǎng)絡遞歸神經(jīng)網(wǎng)絡參考模型回歸正則化強化學習表征學習表示定理再生核希爾伯特空間重采樣法再縮放殘差映射殘差網(wǎng)絡受限玻爾茲曼機限定等距性重賦權(quán)法穩(wěn)健性/魯棒性根結(jié)點規(guī)則引擎規(guī)則學習鞍點樣本空間鞍點樣本空間采樣評分函數(shù)自動駕駛自組織映射半樸素貝葉斯分類器半監(jiān)督學習半監(jiān)督支持向量機SamplespaceSamplingScorefunctionSelf-DrivingSelf-OrganizingMap／SOMSemi-naiveBayesclassifiersSemi-SupervisedLearningsemi-SupervisedSupportVectorMachineSentimentanalysis 情感分析Separatinghyperplane 分離超平面Sigmoidfunction Sigmoid函數(shù)Similaritymeasure 相似度度量Simulatedannealing 模擬退火Simultaneouslocalizationand同步定位與地圖構(gòu)建mappingSingularValueDecomposition 奇異值分解Slackvariables 松弛變量Smoothing 平滑Softmargin 軟間隔Softmarginmaximization 軟間隔最大化Softvoting 軟投票Sparserepresentation 稀疏表征Sparsity 稀疏性Specialization 特化SpectralClustering 譜聚類SpeechRecognition 語音識別

Splittingvariable切分變量Squashingfunction擠壓函數(shù)Stability-plasticitydilemma可塑性-穩(wěn)定性困境Statisticallearning統(tǒng)計學習Statusfeaturefunction狀態(tài)特征函Stochasticgradientdescent隨機梯度下降Stratifiedsampling分層采樣Structuralrisk結(jié)構(gòu)風險Structuralriskminimization/SRM結(jié)構(gòu)風險最小化Subspace子空間Supervisedlearning監(jiān)督學習／有導師學習supportvectorexpansion支持向量展式SupportVectorMachine/SVM支持向量機Surrogatloss替代損失Surrogatefunction替代函數(shù)Symboliclearning符號學習Symbolism符號主義Synset同義詞集LetterTT-DistributionStochasticNeighbourT分布隨機近鄰嵌入Embedding/t-SNETensor 張量TensorProcessingUnits/TPU 張量處理單元Theleastsquaremethod 最小二乘法Threshold 閾值Thresholdlogicunit 閾值邏輯單元Threshold-moving 閾值移動TimeStep 時間步驟Viterbialgorithm Viterbialgorithm 維特比算法TokenizationTrainingerrorTokenizationTrainingerrorTraininginstanceTransductivelearningTransferlearningTreebankTria-by-errorTruenegativeTruepositiveTruePositiveRate/TPRTuringMachineTwice-learningLetterUUnderfittingUndersamplingUnderstandabilityUnequalcostUnit-stepfunctionUnivariatedecisiontreeUnsupervisedlearningUnsupervisedlayer-wisetrainingUpsamplingLetterVVanishingGradientProblemVariationalinferenceVCTheoryVersionspace訓練誤差訓練示例／訓練例直推學習遷移學習樹庫試錯法真負類真正類真正例率圖靈機二次學習欠擬合／欠配欠采樣可理解性非均等代價單位階躍函數(shù)單變量決策樹無監(jiān)督學習／無導師學習無監(jiān)督逐層訓練上采樣梯度消失問題變分推斷VC維理論版本空間Zero-shotlearningZero-shotlearning模式識別計算機硬件的迅速發(fā)展，計算機應用領域的不斷開拓，急切地要求計算機能更有效地感知諸如聲音、文字、圖象、溫度、震動等等信息資料，模式識別便得到迅速發(fā)展。模式VonNeumannarchitectureLetterWWassersteinGAN/WGANWeaklearnerWeightWeightsharingWeightedvotingWithin-classscattermatrixWordembeddingWordsensedisambiguationLetterZZero-datalearning馮?諾伊曼架構(gòu)Wasserstein生成對抗網(wǎng)絡弱學習器權(quán)重權(quán)共享加權(quán)投票法類內(nèi)散度矩陣詞嵌入詞義消歧零數(shù)據(jù)學習零次學習一詞的本意是指模式 一詞的本意是指完美無缺的供模仿的一些標本。模式識別就是指識別出給定物體所模仿的標本。人工智能所研究的模式識別是指用計算機代替人類或幫助人類感知模式，是對人類感知外界功能的模擬，研究的是計算機模式識別系統(tǒng)，也就是使一個計算機系統(tǒng)具有模擬人類通過感官接受外界信息、識別和理解周圍環(huán)境的感知能力。 模式識別是一個不斷發(fā)展的新學科，它的理論基礎和研究范圍也在不斷發(fā)展。隨著生物醫(yī)學對人類大腦的初步認識，模擬人腦構(gòu)造的計算機實驗即人工神經(jīng)網(wǎng)絡方法早在年代末、年代初就已經(jīng)開始。至今，在模式識別領域，神經(jīng)網(wǎng)絡方法已經(jīng)成功地用于手寫字符的識別、汽車牌照的識別、指紋識別、語音識別等方面。目前模式識別學科正處于大發(fā)展的階段，隨著應用范圍的不斷擴大，隨著計算機科學的不斷進步，基于人工神