Artificial Neural Network Applications - 4 Real World ...

2. Artificial Neural Network Applications · Handwriting Recognition – The idea of Handwriting recognition has become very important. · Traveling Salesman Problem ... Skiptocontent MachineLearningTutorials 0 ArtificialNeuralNetworkApplications–4RealWorldApplicationsofANN 1.ANNApplications–Objective Here,wewilldiscuss4real-worldArtificialNeuralNetworkapplications(ANN). TheArtificial NeuralNetworkhasseenanexplosionofinterestoverthelastfewyearsandisbeingsuccessfullyappliedacrossanextraordinaryrangeofproblemdomainsintheareasuchasHandwritingRecognition,Imagecompression,TravellingSalesmanproblem,stockExchangePredictionetc. So,let’sstartApplicationsof ArtificialNeuralNetwork. ArtificialNeuralNetworkApplications–4RealWorldApplicationsofANN Stayupdatedwithlatesttechnologytrends JoinDataFlaironTelegram!! 2.ArtificialNeuralNetworkApplications Here,wewillseethemajorArtificialNeuralNetworkApplications. HandwritingRecognition–TheideaofHandwritingrecognitionhasbecomeveryimportant.ThisisbecausehandhelddeviceslikethePalmPilotarebecomingverypopular.Hence,wecanuseNeuralnetworkstorecognizehandwrittencharacters. TravelingSalesmanProblem–Neuralnetworkscanalsosolvethetravelingsalesmanproblem.Butthisistoacertaindegreeofapproximationonly. ImageCompression–Vastamountsofinformationisreceivedandprocessedatoncebyneuralnetworks.Thismakesthemusefulinimagecompression.WiththeInternetexplosionandmoresitesusingmoreimagesontheirsites,usingneuralnetworksforimagecompressionisworthalook. StockExchangePrediction–Theday-to-daybusinessofthestockmarketisverycomplicated.Manyfactorsweighinwhetheragivenstockwillgoupordownonanygivenday.Thus,Neuralnetworkscanexaminealotofinformationinafastmannerandsortitallout.Sowecanusethemtopredictstockprices. DoyouknowaboutANNModel BelowisthedescriptionofeveryANNapplicationtogettheproperunderstanding. 2.1.HandwritingRecognition Theideaofusingfeedforwardnetworkstorecognizehandwrittencharactersisstraightforward.Thebitmappatternofthehandwrittencharacterisinput,withthecorrectletterordigitasthedesiredoutput.Suchprogramsneedtheusertotrainthenetworkbyprovidingtheprogramwiththeirhandwrittenpatterns. Thetwocommonapplicationsofhandwritingrecognitionare: Opticalcharacterrecognitionfordataentry Validationofsignaturesonabankcheque Feed-forwardnetworkshavethefollowingcharacteristics: a.First,theyarrangeperceptronsinlayers,withthefirstlayertakingininputsandthelastlayerproducingoutputs.Themiddlelayershavenoconnectionwiththeexternalworld,andhencewecallthemhiddenlayers. b.Eachperceptroninonelayerisconnectedtoeveryperceptrononthenextlayer.Henceinformationis“fedforward”fromonelayertothenextinacontinuousmanner.Thisexplainswhywecallthesenetworksfeed-forwardnetworks. c.Thereisnoconnectionamongperceptronsinthesamelayer. Let’sreviseRecurrentNeuralNetworks 2.2.TravelingSalesmanProblem Thetravelingsalesmenproblemreferstothefindingtheshortestpossiblepathtotravelallcitiesinagivenarea.WecanuseNeuralNetworkstosolvethisproblem. Aneuralnetworkalgorithmsuchasageneticalgorithmstartswithrandomorientationofthenetwork,tosolvetheproblem.Thisalgorithmchoosesacityinarandommannereachtimeandfindsthenearestcity.Thus,thisprocesscontinuesseveraltimes.Aftereveryiteration,theshapeofthenetworkchangesandnetworkconvergestoaringaroundallthecities. Theusedalgorithmminimizesthelengthofrings.Inthisway,wecanestimatethetravelingproblem. 2.3.ImageCompression ANeuralNetworkusedforimagecompressioncontaintheequalsizeofinputandoutputlayer.Theintermediatelayerisofsmallersize.Theratiooftheinputlayertotheintermediatelayeristhecompressionratioofthenetwork. Wecangetthecomparisonratioforimagecompressionusingthefollowingformula: ComparisonRatio=InputLayer/IntermediateLayer Ideabehinddatacompressionneuralnetworksistostore,encryptandre-createtheactualimageagain.Thusinsuchnetwork,wecanuseinputfortrainingpurposesitself. Have alookatTopMachineLearningAlgorithm 2.4.StockExchangePrediction Thepredictionaccuracyofneuralnetworkshasmadethemusefulinmakingastockmarketprediction.Forlargebusinesscompanies,makingpredictionsforstockexchangeiscommon.Thisisbyusingparameters,suchascurrenttrends,politicalsituation,publicview,andeconomists’advice. Wecanalsouseneuralnetworksincurrencyprediction,businessfailureprediction,debtriskassessment,andcreditapproval. CompaniessuchasMJFuturesclaimamazing199.2%returnsovera2-yearperiodusingtheirneuralnetworkpredictionmethods. DeanBarrandWalterLoickatLBSCapitalManagementachievedgoodresultsusingasimpleneuralnetwork.Ithad6financialindicatorsonlyasinputs.TheseincludeADX,thecurrentvalueoftheS&P500,andthenetchangeintheS&P500valuefromfivedaysprior.ADXindicatestheaveragedirectionalmovementovertheprevious18days. So,thiswasallabout ArtificialNeuralNetworkApplications.Hopeyoulikeourexplanation. 3.Conclusion TheANNistheveryusefulmodelandtheANNcouldbeappliedinproblem-solvingandmachinelearning.ThecomputingWorldhasalottogainfromtheNeuralNetwork.Thus,Theirabilitytolearnbyexamplemakesthemveryflexibleandpowerful.HencetobestutilizetheANNfordifferentproblems,itisessentialtounderstandthepotentialaswellasthelimitationsoftheNeuralNetwork. If,youhaveanyqueryrelatedtoArtificialNeuralNetworkApplications,feelfreetosharewithus.Wewillbegladtosolvethem. SeeAlso- MachineLearningApplicationsintheRealWorld Forreference IfyouareHappywithDataFlair,donotforgettomakeushappywithyourpositivefeedbackonGoogle|Facebook Tags:ArtificialNeuralnetworkApplicationsmachinelearningMachineLearningtutorialTravellingSalesmanProblem NoResponses Comments1 Pingbacks0 OludareAbiodunsays: August6,2018at7:34am 1.WhataretheArtificialNeuralNetworksapplicationproblemstopatternrecognition? 2.Whataretheupdateinresearchanddevelopmentinresolvingthesechallengesin(1)? 3.WhatareotherfuturechallengesofArtificialNeuralNetworksapplicationproblemstopatternrecognition? 4.WhataretheareasoffutureprospectsinArtificialNeuralNetworksapplicationtopatternrecognition? Reply LeaveaReplyCancelreplyYouremailaddresswillnotbepublished.Requiredfieldsaremarked*Comment*Name* Email* Website ThissiteisprotectedbyreCAPTCHAandtheGooglePrivacyPolicyandTermsofServiceapply. Δ MachineLearningTutorials ML–Introduction ML–Basics ML–Softwares ML–Applications ML–TypesofAlgorithms ML–Classification ML–Tools ML–BestWaytoLearn ML–Future ML–WhyPopular ML–Algorithms ML–UseCases ML–Advantages&Limitations ML–TransferLearning ML–JavaLibraries ML–Clustering ML–GaussianMixtureModel ML–ConvolutionalNeuralNetwork ML–RecurrentNeuralNetwork ML–ArtificialNeuralNetwork ML–ANNApplications ML–ANNLearningRules ML–ANNModel ML–ANNAlgorithms ML–Education ML–Healthcare ML–Finance ML–Entrepreneurs ML–DeepLearning ML–DLTerminologies ML–DLForAudioAnalysis ML–SupportVectorMachine(SVM) ML–SVMApplications ML–SVMKernelFunctions ML–DimensionalityReduction ML–GradientBoostingAlgorithm ML–XGBoostIntroduction ML–XGBoostingAlgorithm ML–AdaBoostAlgorithm DeepLearningvsML DeepLearningvsMLvsAIvsDS HowGoogleusesMachineLearning MachineLearningCaseStudies MLInfographicforBeginners MachineLearningProjects MachineLearningProjectIdeas ML–Projects 70+ProjectIdeas&Datasets MachineLearningProject–CreditCardFraudDetection MachineLearningProject–SentimentAnalysis MachineLearningProject–MovieRecommendationSystem MachineLearningProject–CustomerSegmentation MachineLearningProject–UberDataAnalysis DeepLearningProjectIdeas PythonMLTutorials PythonML–Tutorial PythonML–EnvironmentSetup PythonML–DataPreprocessing PythonML–Train&TestSet PythonML–Techniques PythonML–Applications PythonML–Algorithms PythonDeepLearningTutorial PythonDL–Applications PythonDL–EnvironmentSetup PythonDL–PythonLibraries PythonDL–NeuralNetwork PythonDL–ComputationalGraphs