Issuu

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p ISSN: 2395 0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page1014 A Study on Machine Learning Techniques for Predicting Software Defects Akhilesh G.V Dhavileswarapu1, Namburi Rama Bhadra Raju2, Rimmalapudi Sri Nihaal3 , Shanmuk Srinivas Amiripalli4 1,2,3,4Department of CSE, GIT, GITAM UNIVERSITY, Visakhapatna, AP, India. *** Abstract High quality software cannotbedevelopedwithout resorting to high quality o testing. Testing object oriented programs, the object oriented features need to be tested specifically. Software defect prediction is especially useful for testing of large software systems. Software defect prediction and its techniques used to at present for testing object oriented software pay very little attention to adequately test object oriented features. Ourpreliminaryreviewandproposed model show that, using proper SDP mutant score can be accomplished. Software defect prediction and its techniques used to at present for testing objectorientedsoftwarepay very little attention to adequately test object oriented features. Key Words: software engineering, software defect, ML Techniques, Complexity based Oversampling Technique, softwaredefectprediction. 1. INTRODUCTION mawhetherdataSDPandequal,defectiveClassSDP,detectpatternsemployingsituations.[6]developmentloweringsoftwaresoftwareinadequacy,Thesesimpleasintelligence.amongdefectsoftware.wasUntiluntil,testinganddebuggingsoftwareonaregularbasisanecessarystepindevelopinghighlydependableAdeeplearningapproachknownassoftwareprediction(SDP)hasgrownincreasinglypopularengineerswiththeintroductionofartificialSDPmodelsidentifydefectpronethings(suchfunctionsandfiles)[3].SDPmakestestinganddebuggingandeffectivefordevelopersbyremovingobstacles.softwaredefectsarecausedbyasoftwaresystem'swhichleadstounexpectedconditionsaftertheissentout,resultinginasituationwheretheproductdoesnotmeettheclient'spreferences,thesoftwareproductqualityandincreasingcosts.AproductivesoftwarepredictionmodelmayhelpdevelopersandmaintainersavoidsuchTheempiricalfindingsalsosuggestthatsentencebasedandkeywordbasedpredictionmayhelppretrainedneurallanguagemodelssoftwaredefectsbetter[2].ontheotherhand,hasissueswithclassimbalance[5].imbalanceisanissuethatoccurswhentheamountofandnondefectiveoccurrencesinadatasetarenotcausingpredictionmodelstofavourthemajorityclassdisregardtheminorityclass[3].Toaddressthisissue,employsmachinelearningmethodsthatutiliseprevioustobuilddefectpredictionmodelsthatcanpredictfutureinstanceswillbefaultyornot.Becausechinelanguageapplicationsincludeseveralrealworld simulations,machine learningmethodsareseen to bethe ToidealsolutiontoSDP.copewiththeissue of class imbalance in machine andMinorityapproachapproachnearestinstancesinstancespredictioninstanceshighertheproposedOversamplingOversamplingOversamplingbringhand,samplinginformationdominantInstrategiesonareThelearning,datasamplingapproachesareused.oversamplingmethodandtheundersamplingapproachtwokindsofdatasamplingprocedures.Byconcentratingnormalisingthedistributionofthedatasets,bothoftheseincreasethepredictionmodel'seffectiveness[3].SDP,however,theoversamplingmethodhasbeenthemethod.Thereisariskoflosingcrucialandusefulforthepredictionmodelwhenusingtheunderstrategy.Oversamplingstrategies,ontheotheraddanewinstancetotheminorityclassinordertoitsvalueclosertoorequaltothatofthemajorityclass.approachessuchasCOSTE(ComplexitybasedTechnique)andSMOTE(SyntheticMinorityTechnique)havebeendeveloped.Ithasbeenbyacademicsandseveralpeerstudiestoaddressissueofclassimbalance[3,15].TheCOSTEtechniqueusesrankedinstancesmoreoftenthanlowerrankedtogeneratemorebalanceddatasets,causingthemodeltopaymoreattentiontothelesscomplexandthenaveragingthehigherandlowerrankedtogethertogeneratesyntheticinstances.TheKneighbour(KNN)methodisusedintheSMOTEtocreateasyntheticminorityclass.AnotherbasedonSMOTEisS.M.O.T.U.N.E.D(SyntheticOversamplingTUNED),whichisaneuralnetworkalgorithmusedtoanalyseclassimbalances. 2. ML TECHNIQUES AND LITERATURE SURVEY softwareprofestimatornearesthidden(J48),datasets.usedamongcomponentswayMLapproachesareregardedasanoperativeandoperationalforlocatingproblematicmodules,inwhichmovingarediscoveredbymininghiddenpatternssoftwaremeasurements.MLapproachesarebeingbyanumberofacademicsworkingwithhealthcareThemultilayerperceptron(MLP),decisiontreeradialbasisfunction(RBF),randomforest(RF),Markovmodel(HMM),credaldecisiontree(CDT),Kneighbour(KNN),averageonedependency(A1DE)[4],andNaveBayes(NB)[2025]aresomethemachinelearningapproaches.Techniquesforedictingsoftwarefaultsaidintheidentificationofsystemcomponentsthataremorelikelytohave

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p ISSN: 2395 0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page1015 flaws. Models may be built using defect prediction papers'theoverview[13].machinelearningnormallearningdetectionsamplesinformedSyntheticmachinefromdiscriminativedefecttraceability,linkhaveprocesses(e.g.,SomeresultingapproachesachieveuniqueCOSTEallBoosting,[1].numberapproachestoranksoftwaremodulesbasedontheprojectedoffaults,defectlikelihood,orclassificationfindingsSamplingStrategies,CostsensitiveLearningTechniques,andCrossProjectSoftwareDefectPredictionareMLtechniquesaddressingclassimbalanceproblems[2].(ComplexitybasedOversamplingTechnique)isanoversamplingtechniquethatcanconcurrentlylowpfandhighpd[3].Softwaredefectpredictionautomaticallydetectprobableproblems,inconsiderabletime,effort,andcostsavings[19].researchershavealreadyuseddeeplearningmethodsCNNandDBN)toenhancesoftwareengineeringinrecentyears[2631].Deeplearningapproachesalsobeenappliedinthecategorizationoftestreports,predictioninonlinedeveloperforums,softwareandotherapplications[6].Themajorityofpredictionalgorithmsrelyonmanuallycreatingnewfeaturesornovelcombinationsoffeatureslabelledhistoricaldefectdata,whicharethenfedintolearningbasedclassifierstodetectcodeflaws[11].minorityoversamplingmethods(SMOTE),likeoversampling,producesyntheticminorityclasstobalancetheclassdistribution[14].AnomalybasedNIDShavewidelyemployedmachineapproaches[3237].Todiscriminatebetweenandaberrantnetworkactivity,manymachinemodelshavebeenused,includingsupportvector(SVM),randomforest(RF),anddecisiontree(DT)Thetableinthissectionofthestudyprovidesanofsomeoftheresearchpublicationspublishedinlastfewyears.Thefollowingtablesummarisesthetechniques,limits,andbenefits. 3. SOFTWARE TOOLS Following are some of the important software defect whichpredictiontoolswhichareavailablecommerciallyandsomeareavailableforuseforfree[3842]. NAME DECRIPTION Class LearningImbalance Class imbalance learning focuses on classifying issues with skewed distributions, which might be useful fordefectprediction. NetworksBayesian To discover the important probabilistic correlations among Bayesiansoftwaremetricsandfaultproneness,networkswereused. NeuralConvolutionalNetworks Deep learning is used to generate tokeneffectivefeatures.Weinitiallyextractvectorsfromtheprogrammes' Abstract Syntax Trees (ASTs), which embedding.througharethenencodedasnumericalvectorsmappingandword Deep CodeModelLearningonStaticFeatures It's significant because it shows a fresh application of deep learning models to an actual challenge encounteredbysoftwareengineers. DeepTree Based Model It can learn characteristics for modelling source code automatically Treeitandusethemtoforecastdefects,anddirectlymatchestheAbstractSyntaxrepresentationofsourcecode. 4. REPOSITORIES FOR SOFTWARE DEFECT PREDICTION The following table contains the information of the softwarerepositoriesthatwefound,whichcontainsdatasetsforthedefectprediction. oryRepositofName Description URL TYPE Github1 Xiang Chen maintains this repository. mlse.github.io/sdp.hthttps://smart Kaggle Kaggle is an GooglesubordinateKagglepractitioners.machinedatascientistscommunityonlineofandlearningisaofLLC.ct+prediction?q=software+defekaggle.com/searchhttps://www. PUBLIC sDatasetGoogle Data Repository byGOOGLE KS5shorturl.at/pG PUBLIC vData.Go DR0shorturl.at/st PUBLIC Promise RepositorySoftwareEngineering etsERepository/datase.site.uottawa.ca/Shttp://promispage.html PUBLIC rldData.Wo shorturl.at/ju yC6 PUBLIC IITKGP National Digital libraryofindia Iitkgp.ac.in NALINSTITUTIO AUTHOR METHODOLOGY LIMITATIONS CONCLUSIONS ADVANTAGES Panetal. CodeBERT NT, CodeBERT PS, CodeBERT PK, and CodeBERT PT are some of the CodeBERT models 1. Language for programming. The PROMISE repository, which is developed in These among the CodeBERT models proposedinthisstudy for software defect The use of a pre trained CodeBERT model enhances prediction performance and saves

prediction.proposedforsoftwarefault

projectversionusingCodeBERTdefectpatternsvarioustheperformance,improveCodeBERTlanguageifprediction.Theytestedutilisinganeuralmodellikemightpredictionaswellasconsequencesofpredictioninsoftwarepredictionusingmodels,modelsincrossandcrossSDP. time, according to empiricalfindings.a

Thegoalofthisresearchis to look at the stability of oversampling approaches based on SMOTE. Furthermore, to increase the stability of SMOTE based oversampling methods, a series of stable SMOTE oversamplingbased techniques arepresented. The fault measurements

Fengetal.

sortsittheBecausemightusedinthisinvestigationbeadrawback.theyonlyusedstaticcodemeasure,can'tbeusedtootherofmetrics.

OSTE).nghniquexityTheysuggestedtheComplebasedOverSamplingTEcasanewoversampliapproachinthiswork(C

Java, provided this dataset. The strategy is requirement.aproject'sinaccessibleNASAdatasets,pastdataset,isthecollection2.programmingnot,however,confinedtolanguages.ThesizeofthedataAportionofPROMISErepositoryemployedinthisasithasbeeninstudies.Somesuchasthedataset,areduetotheopensource

Fengetal.

The findings of this experiment, which used 23 datasets, cannot be appliedtootherkindsof metrics, such as process metrics. The findings' external validity is jeopardised due to their measures.otherlackofgeneralizabilitytodatasetsor

balanceveaCOSTEissuggestedasviableapproachtosoltheissueofclassiminSDP. COSTE considerably enhances the variety of synthetic instances without reducing the effectiveness of prediction models to discover errors, according to experimental data from 23 releases of ten projects.

gOTEasOTEhesOversamplingapproacbasedonstableSMshouldbeviewedanalternativeforSMbasedoversamplintechniques. This comparison of SMOTE basedandstable SMOTE oversamplingbased strategies reveals that the stable SMOTE oversamplingbased technique outperformstheformer. Zhuetal predictoramachinekernelnetworkconvolutional(SA).searchdevelopedsimulatedalgorithmwhaleUsingtherecentlyreleasedoptimization(WOA)andannealing,theyEMWS,afeaturebasedtechniqueTheyalsousedaneural(CNN)andaextremelearningtocreateWSHCKE,unifieddefectprediction(KELM). Even if we thoroughly examine the experiment method, there may still widespreaddataset'sAnotherunnoticed.besomemistakesthatgoissueisthequalityandavailability. weakcapabilitystrongleveragingperformanceexploitationWOA'scapabilitystrongproject,forrepresentativecloselythatselectionmetaheuristicimprovedTheycreatedEMWS,anfeaturealgorithmselectsfewerbutrelatedfeatureseachsoftwareleveragingSA'slocalsearchtoimproveweakwhilealsoWOA'sglobalsearchtoboostSA'sexploration. The EMWS developed here may efficiently predictionfeatures,abstractcharacteristicscombineWSHCKEcloselycharacteristicsofchooseasmallernumberrepresentativethatareconnected.mayuseCNNtothespecifiedintodeepsemanticimprovingperformance. etMaddipatial. The most significant qualitiesindetectingfaulty When compared to classifierssuchasNeural To tackle the issue of class imbalance, a Detecting software components that are

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p ISSN: 2395 0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page1017 prone modules were identified using principal component analysis as an attribute selection technique. Networks and Support Vector Machines, the Sugeno Fuzzy Inference classifier increased the byareaundertheROCcurvejust5%. machine learning method is being used to predict software defects. pronetofailure etSahingozal. Error detection and error correction. Thedefectiveclasshasa thangreaterclassificationcosttheinertclass. The presented models provide appropriate accuracy levels for software defect quality.higherprediction,resultinginsoftware Not only can the programme be used on normal desktop sensorcomputingbecomputers,butitcanalsousedontinydeviceslikenetworks. Esteves et al. Uses a tree boosting technique that takes a training set of records of easy to computeproperties modulereturnsofeachmoduleasinputandwhetherthatisdefectprone. The majority of these studies focus on predicting errors from a becomewhatofliteraturemajorcharacteristics.largenumberofsoftwareAnotherflawinthepresentistheabsenceagoodexplanationofcausessoftwaretoflawed. We give particular developers.resultsprojects,defectivenessimpactsoftwareelementsthattheofchosenthusthearevaluableto The less features aid model explainability, which is vital for defects.whichmodconnecteddevelopersprovidinginformationtoonfeaturestoeachuleofthecode,ismoreproneto Panetal. By adaptively selecting characteristics.data'sstrengtheningexpandingtheselectedminoritytheSMOTEmethodclass,datagroupsofInnerandDangerfromtheminoritytheAdaptiveSMOTEimprovesthemethod,resultingincreationofanewclassbasedonthedata,preventingcategoryboundaryfromandtheoriginaldistributional The distribution of the original minority data is not well grasped by samplingshortcwhichfreshlybalancedsamples,isamajoromingofcurrentapproaches.

The intrusion detection dataset's unbalanced class issue restricts the classifier's performance forminorityclasses. SGM CNN beats state of the art intrusion detection approaches and offers an effective intrusionsolutiontounbalanceddetection. Ability to mine sensitive data on the distinctions between normal and malignantactivity.

Singh et al. To successfully solve unbalanced classification issues, this paper introduces synthetic minority over sampling strategies based on class specific extreme learning machines. By developing fresh synthetic samples usingtheSMOTEapproach, For data with a lot of overlapmaysyntheticaccountdifferentnearbySMOTEparticularlydimensions,SMOTEisn'tuseful.doesnottakeinstancesfromclassesintowhenproducingexamples.Thisleadtomoreclassandmorenoise.

characteristics.distribution

on the

Zhang et al. SGM, a novel class imbalance processing approach that combines Synthetic Minority Over Sampling Technique (SMOTE) with under sampling for clustering largeModel,basedonGaussianMixtureisavailableforscaledatasets(GMM).

To deal with skewed class distribution and difficulties.Thisclassificationunbalanced paper SMOTE.Bybasedlearningaproposesandevaluatesclassspecificextrememachineonproducing This study introduces a novel SMOTE based class specific extreme learning machine, a variant of the class soversamplingemploysmachinespecificextremelearning(CSELM),thatbothminorityandclasspecificregularisation.

features.

data

Adaptive SMOTE divides the positive dataset into Danger and Inner and then oversamples the using SMOTE depending original data This method avoids the spread of positive and the original dataset's distributional divisionanGaussianOversamplingisuniquesamplingapproach.

improves

samples

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p ISSN: 2395 0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page1018 the number of samples belonging to the minority classrises. synthetic samples belonging to the minority class, SMOTE raisestherelevanceof the minority class samples for defining classifiers.thedecisionareaofthe The synthetic minority oversampling technique regiondeterminingclassthe(SMOTE).minoritywasusedinthisworkforoversamplingItemphasisesrelevanceofminoritysamplesinthedecisionofclassifiers. Chenetal. They introduced an adaptive robust SMOTE, dubbed RSMOTE, for class imbalance classification in thisstudy. It's different from most current approaches, andwhichmakesitintriguingunreliable. For parameters.orparticulardoesotherprapproachnoise,classificationunbalancedwithlabeltheRSMOTEisesented.Unlikemostapproaches,itnotdependonanoisefilteraddanyadditional RSMOTE promotes byminimiseszonesthesamplessafetheadaptivelyclassreduceslearninginthreeways:itbiascausedbyimbalance;itselfdiscriminatesnoisy,borderline,andsortsofminorityandreinforcesboundaryandsafetyseparately;anditbiasinducedclassimbalance. Douzas et al. To rebalance skewed datasets, the approach suggested in this paper withalgorithmkusesthebasicandcommonmeansclusteringincombinationSMOTEoversampling. moretoexamples.producingdifferentthatsurroundingSMOTEdoesnotexamineinstancesmightbefromclasseswhilesyntheticThismayleadmoreclassoverlapandnoise. The suggested characteristicsaccomplishestechnique by space.partsfocuseddatameans,clusteringdatausingkwhichallowsproductiontobeoncriticaloftheinput This paper provides a simple and efficient oversampling approach based on k means clustering and SMOTE (synthetic minority solvescreationwhichoversamplingtechnique),preventsnoiseandsuccessfullyclassimbalances. PrabhaLakshmiC. For the prediction of software flaws, a hybrid stratedependentandfeaturereductionapproachanartificiallyneuralnetworkgyareprovided. Data standardisation is required for the PCA approach employed in this investigation.Your initial features will become Principal PrincipalComponents.andfeaturescomponents.makesyourTheappliedComponentswhenPCAistothedataset.linearcombinationoforiginalattributesupthemainOriginalaremorelegibleinterpretablethan predictionwithoutpredictiondifficultiesperioddiscoveredstrategies,featuresoftwaremethodstoThegoalofthisstudyisemploydataminingtoforecastfaults.Usingselectionithasbeenthattheandspacefordefectisreducedreducingaccuracy. According to the research, the suggested ofoversignificant98.70%,producestechniqueisefficientandanAUCofwhichisaimprovementvariousstatemodelsthepast.z Rahim et al. The assumption of independent predictor traits is Naive Bayes' primary flaw. All of the independent.assumedqualitiesinNaiveBayesaretobemutually reductionstagesincludessections,downframeworktrustworthy.bsoftwareparadigmTheresearchdevelopedaforpredictingproblemsthatisothefficientandTheisbrokenintothreeprimaryeachofwhichdatapreparationsuchas:noiseand The suggested results.accordingBayesutilisingatechniquemayachieve98.7%accuracytheNavealgorithm,tothe result,suchdeveloperssystems,earlybyreduceofthattechnique'sAsaresult,thesuggestedrelevanceisitwouldcutthecostmaintenanceandcodecomplexityanticipatingflawsinsoftwarewhichwillaidinremovingproblemsand,asaenhancesoftware

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p ISSN: 2395 0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page1019 normalisation quality prior to deployment. PROPOSED MODEL Fig 1:proposedmodel 5. CONCLUSIONS detectingpredictionevaluatedobjectincreases.quantitysoftwarewherethemaySoftwaredefectpredictionhasgainedalotoftraction,andithelpyousavemoneyandtimeonyourproject.Someoffundamentalaspectsofobjectorientedprogrammingarethebiggestcomplicationsoccur.TherelevanceofdefectpredictionissteadilyincreasingastheandcomplexityofcontemporarysoftwareproductsRecentprogrammingmethodologies,suchasorientedsystems,cannot,however,bethoroughlyonlybylookingatthesourcecode.Defectusingmachinelearningisapotentialmethodforsoftwareprojectflaws.. ACKNOWLEDGEMENT UniversityDepartmentTheauthorswouldliketothankprofSrinivasPrasadandtheofComputerScienceEngineeringatGITAMfortheirassistance. REFERENCES [1] Lei Qiao, Xuesong Li, QasimUmer, Ping Guo; Deep learningbasedsoftwaredefectprediction;Qiao, Lei,et al. "Deep learning based software defect prediction." Neurocomputing385(2020):100 110. [2] Cong,MinyanLu,andBiaoXu."AnEmpiricalStudyon Software Defect Prediction Using CodeBERT Model." AppliedSciences11.11(2021):4793. [3] Feng, Shuo, et al. "COSTE: Complexity based OverSamplingTEchniquetoalleviatetheclassimbalance Softwareprobleminsoftwaredefectprediction."InformationandTechnology129(2021):106432. [4] Bilal Khan,1RashidNaseem,2MuhammadArifShah,2 (2021).Techniques."BigMahmoud6KarzanWakil,3AtifKhan,4M.IrfanUddin,5andMarwan;"SoftwareDefectPredictionforHealthcareData:AnEmpiricalEvaluationofMachineLearningJournalofHealthcareEngineering2021

[5] Feng, Shuo, et al. "Investigation on the stability of SMOTE based oversampling techniques in software defect prediction." Information and Software Technology(2021):106662.

[9] Cetiner, Murat, and OzgurKoraySahingoz. "A Comparative Analysis for Machine Learning based Software Defect Prediction Systems." 2020 11th

[6] Zhu, Kun, et al. "Software defect prediction based on andandenhancedmetaheuristicfeatureselectionoptimizationahybriddeepneuralnetwork."JournalofSystemsSoftware(2021):111026.

[7] TechnologypredictionbiasedYao,Jingxiu,andMartinShepperd."Theimpactofusingperformancemetricsonsoftwaredefectresearch."InformationandSoftware(2021):106664.

[8] Maddipati, Satya Srinivas, and Malladi Srinivas. "Machine learning approach for classification from MaterialsimbalancedsoftwaredefectdatausingPCA&CSANFIS."Today:Proceedings(2021).

[27]

[20] Amiripalli, S. S., Kollu, V. V. R., Jaidhan, B. J., Srinivasa Chakravarthi, L., & Raju, V. A. (2020). Performance improvement model for airlines connectivity system using network science. International Journal of 9(1),AdvancedTrendsinComputerScienceandEngineering,789792.

Conference on Applied Sciences and Technologies (IBCAST).IEEE,2021. [19] (2021):InnovationspredictlearningKumar,P.Suresh,etal."Bootstrapaggregationensemblebasedreliableapproachforsoftwaredefectionbyusingcharacterizedcodefeature."inSystemsandSoftwareEngineering125.

[21] Amiripalli,S.S.,&Bobba,V.(2020).AFibonaccibased Research,InternationalTGOmethodologyforsurvivabilityinZigBeetopologies.JournalOfScientific&Technology9(2),878881.

[25] Potharaju,S.P.,Sreedevi,M.,&Amiripalli,S.S.(2019). An ensemble feature selection framework of sonar targetsusingsymmetricaluncertaintyandmulti layer perceptron(su mlp).InCognitiveInformaticsandSoft Computing(pp.247 256).Springer,Singapore.

ControlSystemP.Jitendra,M.S.,Amiripalli,S.S.,Kollu,V.V.R.,Chowdary,R.,&Rao,R.V.(2020).AnalysisofAirlineConnectivityusingGraphTheory.InternationalJournalofandAutomation,13(4),7784.

[22] Amiripalli,S.S.,Bobba,V.,& Potharaju,S.P.(2019).A novel trimet graph optimization (TGO) topology for wireless networks. In Cognitive Informatics and Soft Computing(pp.75 82).Springer,Singapore.

[28] Jitendra, M. S., Srinivasu, P. N., Shanmuk Srinivas, A., 1236machineonNithya,A.,&Kandulapati,S.K.(2020).Crackdetectionconcreteimagesusingclassificationtechniquesinlearning.JournalofCriticalReviews,7(9),1241.

[29] 79(4),ScientificbasedAmiripalli,S.S.,&Bobba,V.(2020).AnOptimalGraphZigBeeMeshforSmartHomes.JournalofandIndustrialResearch,ScientificPublishers,318322.

[32] T.Kollu,V.V.,Amiripalli,S.S.,Jitendra,M.S.N.V.,&Kumar,R.(2021).ANetworkScienceBasedPerformance

with Naïve

[31] Jitendra,M.S.,Srinivas,A.S.,Surendra,T.,Rao,R.V.,& Chowdary, P. R. (2021, October). A study on game development using unity engine. In AIP Conference Proceedings (Vol. 2375, No. 1, p. 040001). AIP PublishingLLC.

[18] Rahim, Aqsa, et al. "Software Defect Prediction Bayes Classifier." 2021 International Bhurban

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p ISSN: 2395 0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page1020 andInternationalConferenceonComputing,CommunicationNetworkingTechnologies(ICCCNT).IEEE,2020. [10] Esteves, Geanderson, et al. "Understanding machine learning software defect predictions." Automated SoftwareEngineering27.3(2020):369 392. [11] Li, Jian, et al. "Software defect prediction via (QRS).Conferenceconvolutionalneuralnetwork."2017IEEEInternationalonSoftwareQuality,ReliabilityandSecurityIEEE,2017. [12] SciencesonPan,Tingting,etal."LearningimbalanceddatasetsbasedSMOTEandGaussiandistribution."Information512(2020):12141233. [13] ComputerfornetworkZhang,Hongpo,etal."AneffectiveconvolutionalneuralbasedonSMOTEandGaussianmixturemodelintrusiondetectioninimbalanceddataset."Networks177(2020):107315. [14] Raghuwanshi, Bhagat Singh, and Sanyam Shukla. "SMOTEbasedclass specificextremelearningmachine for imbalanced learning." Knowledge Based Systems 187(2020):104814. [15] Chen, Baiyun, et al. "RSMOTE: A self adaptive robust SMOTE for imbalanced problems with label noise." InformationSciences553(2021):397 428. [16] Douzas, Georgios, Fernando Bacao, and Felix Last. "Improving imbalanced learning through a heuristic oversamplingmethodbasedonk meansandSMOTE." InformationSciences465(2018):1 20. [17] InformaticsInternationalpredictionPrabha,C.Lakshmi,andN.Shivakumar."Softwaredefectusingmachinelearningtechniques."20204thConferenceonTrendsinElectronicsand(ICOEI)(48184).IEEE,2020.

[30] Naidu, J. L., Gorakala, A. C., & Amiripalli, S. S. (2020). Hash functions and its security for Snags.IRJET,7(7),3465 3471.

[26] Thota, J. R., Kothuru, M., & Shanmuk Srinivas, A. Monitoring Diabetes Occurrence Probability Using ClassificationTechniqueWithAUI.

[24] Amiripalli, S. S., Venkatarao, R., Jitendra, M. S. N. V., & IntandMycherla,N.M.J.(2020).Detectingemotionsofstudentassessingtheperformancebyusingdeeplearning.JAdvTrendsComputSciEng,9(2),16411645.

[23] Amiripalli, S. S., & Bobba, V. (2019). Trimet graph optimization (TGO) based methodology for scalability and survivability in wireless networks. International Journal of Advanced Trends in Computer Science and Engineering,8(6),3454 3460.

International Research Journal of Engineering and Technology (IRJET) e ISSN: 2395 0056 Volume: 08 Issue: 12 | Dec 2021 www.irjet.net p ISSN: 2395 0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page1021 Improvement Model for the Airline Industry Using NetworkX. International Journal of Sensors Wireless CommunicationsandControl,11(7),768 773. [33]

[41] Amiripalli, S. S., & Bobba, V. (2018). Research on network design and analysis of TGO topology. International Journal of Networking and Virtual Organisations,19(1),72 86.

[42] Amiripalli, S. S., & Bobba, V. (2019). Impact of trimet graph optimization topology on scalable networks. Journal of Intelligent & Fuzzy Systems, 36(3), 2431 2442.

ComInternationalConnectedDesignKollu,V.V.,Amiripalli,S.S.,&Jitendra,M.S.N.V.(2021).andImplementationofanOptimalNEdgeNetworksforIOTBasedRandomMesh.JournalofSensorsWirelessmunicationsandControl,11(7),782788. inImageTechniqueP.,Kanakaraju,R.,Lakshmi,V.,Amiripalli,S.S.,Potharaju,S.&Chandrasekhar,R.(2021).AnImageEncryptionBasedonLogisticSineMapandanEncryptedRetrievalUsingDCTFrequency.InRecentTrendsIntensiveComputing(pp.18).IOSPress. (pp.Conferencefor&Amiripalli,S.S.,Dora,T.,Addagarla,S.K.,Srinivasu,P.N.,Rao,G.S.(2021).AGraphInvariantBasedTGOModelRailTelOpticalNetworks.InSixthInternationalonIntelligentComputingandApplications6979).Springer,Singapore. provingv.talari,s.,amiripalli,s.,sirisha,p.,kumar,d.s.,&deepika,k.animprovedcipherbasedautomatictheoremtechniqueforencryptionanddecryption. L. P. N., NORWAWİ, N., AMİRİPALLİ, S. S., & P. Secured

Application of Big Data Analytics and Pattern Recognition Aggregated With Random Forest for DetectingFraudulentCreditCardTransactions(CCFD BPRRF). [38] SRİNİVASU,

[39] Amiripalli, S. S., Kumar, A. K., & Tulasi, B. (2016, February). Introduction to TRIMET along with its properties and scope. In AIP Conference Proceedings (Vol.1705,No.1,p.020032).AIPPublishingLLC.

[34]

[37] Jaidhan, B. J., Madhuri, B. D., Pushpa, K., & Devi, B.

DEEPALAKSHMİ,

[40] Amiripalli, S. S., & Bobba, V. (2019). An optimal TGO Communications,intopologymethodforascalableandsurvivablenetworkIOTcommunicationtechnology.WirelessPersonal107(2),10191040.

[35]

[36]

Compression for 2D Medical Images Through the Manifold and Fuzzy Trapezoidal Correlation Function. Gazi University JournalofScience.