毕业设计_英文翻译

《毕业设计_英文翻译》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、9Continuous-TimeDynamicNeuralNetworks9连续时间动态神经网络9.1DynamicNeuralNetworkStructures:AnIntroduction9.2HopfieldDynamicNeuralNetwork(DNN)andItsImplementation9.3HopfieldDynamicNeuralNetworks(DNNs)asGradient-likeSystems9.4ModificationsofHopfieldDynamicNeuralNetworks9.5OtherDNNModels9.6ConditionsforE

2、quilibriumPointsinDNN9.7ConcludingRemarksProblems9.1动态神经网络结构导论9.2动态的Hopfield神经网络（DNN）及其实现9.3动态的Hopfield神经网络（DNNS）为梯度样系统9.4修改动态的Hopfield神经网络9.5其他DNN模型9.6条件平衡点在DNN9.7结语问题Asseeninthepreviouschapters,aneuralnetworkconsistsofmanyinterconnectedsimpleprocessingunits,calledneurons,whichformthelayered

3、configurations.Anindividualneuronaggregatesitsweighedinputsandyieldsanoutputthroughanonlinearactivationfunctionwithathreshold.Inartificialneuralnetworkstherearethreetypesofconnections:intralayer,interlayer,andrecurrentconnections.Theintralayerconnections,whicharealsocalledlateralconnectionsor

4、cross-layerconnections,arelinksbetweenneuronsinthesamelayerofthenetwork.Theinterlayerconnectionsarelinksbetweenneuronsindifferentlayers.Therecurrentconnectionsprovideself-feedbacklinkstotheneurons.Ininterlayerconnections,thesignalsaretransformedinoneofthetwoways:eitherfeedforwardorfeedback.就像

5、在前面的章节中，神经网络由许多互连简单处理单元，称为神经元，形成层状结构。单个神经元聚集体的称重输入，并通过与阈值的非线性激活函数产生一个输出。人工神经网络有三种类型的连接：层内，层间，并反复发作的连接。层内的连接，这也被称为横向连接或跨层的连接，在网络的相同层的神经元之间的链接。层间连接是在不同层的神经元之间的联系。经常性的连接提供自反馈链接到神经元。在中间层的连接，信号转换的两种方法之一：要么前馈和反馈。Fromthecomputationalpointofview,adynamicneuralstructurethatcontainsthestatefeedbackmayp

6、rovidemorecomputationaladvantagesthanastaticneuralstructure,whichcontainsonlyafeedforwardneuralstructure.Ingeneral,asmallfeedbacksystemisequivalenttoalargeandpossiblyinfinitefeedforwardsystem(HushandHome1993).Awell-knownexampleisthataninfinitenumberoffeedforwardlogicgatesarerequiredtoemulatea

7、narbitraryfinite-statemachine.Alsoaninfiniteorderfiniteimpulseresponse(FIR)filterisrequiredtoemulateasingle-poleinfiniteimpulseresponse(IIR).Anonlineardynamicrecurrentneuralnetworkstructureisparticularlyappropriateforsystemidentification,cont