Real Life Applications of Soft Computing 4

《Real Life Applications of Soft Computing 4》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、2ArtificialNeuralNetworksI2.1artiFiCialneuralnetWorKSAswehavenoted,aglimpseintothenaturalworldrevealsthatevenasmallchildisabletodonumeroustasksatonce.Themostcomplexproblemsthatengineerskeepstrugglingwitharesosimpleforachildtodo.Letuslookagainattheexamp

2、leofachildwalking.Probablythefirsttimethatchildseesanobstacle,he/shemaynotknowwhattodo.Butafterward,wheneverhe/shemeetsobstacles,he/shesimplytakesanotherroute.Itisnaturalforpeopletobothappreciatetheseobser-vationsandlearnfromthem.Anintensivestudybypeop

3、lecomingfrommultidisciplinaryfieldsmarkedtheevolutionofwhatwecalltheartificialneuralnetwork(ANN).OneofthemostimportantfactorsthathasresultedinahighindustrialuseofANNsistheirabilitytosolveverycomplexproblemswithgreatease.Researchersanddeveloperscanuseth

4、eirknowledgeandexperiencetodesignasimplesystemthatisabletoperformwonders.AlthoughthedesignandimplementationofANNsissimple,agreatdealofeffortmayberequiredtocollectthenecessaryhistoricaldatausedforlearning.Itmayalsobenecessaryfortheresearchersanddevel-op

5、erstoconstantlytrynewerandnewerarchitecturesuntiltheyultimatelyfindanarchitecturethatsolvestheproblemwiththebestresults.ANNshaveagreatabilitytolearnlargevolumesofdata,whichenablesthemtogiveaccurateresultswheneverthedatareappearintheproblem.Thisabilityt

6、olearnismainlyduetotheinherentpropertyofANNsbeingabletoimitatealmostanyfunction.Hencetheyaregreattoolsformachinelearning.Anotherimportantadvantageoftheneuralnetworksistheirabilitytogivethecorrectanswerstounknowninputs.Inmostcases,theyareabletofindtheco

7、rrectanswertounknowninputswithsufficientaccuracieswheneverthatarepresentedtothesystem.Agreatdealofworkoverthepastfewdecadeshasseenverypromisingarchitectures,models,andmethodsforsolvingtheproblemsthatcanbesolvedbytheuseofANNs.Thesenetworkshavebecomevery

8、handytoolsforindustrialimplementations.Infact,ANNsarebeingusedinalmosteveryindustrytosolvereallifeproblems.ManyofthesystemswediscussinthenextunitsuseANNstofindtheoutputstoappliedinputs.ANNsrepresentahighlyconnectednetworkofneurons,theba