ThisarticledescribeshowtheenormoussizeofdatainIoTneedsefficientdataminingmodelfor informationextraction,classificationandmininghiddenpatternsfromdata.CBRisalearning,mining andproblem-solvingapproachwhichsolvesaproblembyrelatingpastsimilarsolvedproblems.One issuewithCBRisfeatureweighttomeasurethesimilarityamongcasestominesimilarpastcases. NN’spruningisapopularmethod,whichextractsfeatureweightsfromatrainedneuralnetworkwithout losingmuchgeneralityofthetrainingsetbyusingfourmechanisms:sensitivity,activity,saliency andrelevance.However,trainingNNwithimbalanceddataleadstheclassifiertogetbiasedtowards themajorityclass.Therefore,thisarticleproposesahybridCBRmodelwithRUSandcostsensitive backpropagationneuralnetworkinIoTenvironmenttodealwiththefeatureweightingproblemin imbalancedata.Theproposedmodelisvalidatedwithsixreal-lifedatasets.Theexperimentalresults showthattheproposedmodelisbetterthanotherfeatureweightingmethods. KEywORdS Artificial Neural Network (ANN), Case Based Reasoning (CBR), Class Imbalance, Data Mining, Internet of Things (IoT)

中文翻译：

---

物联网（IoT）环境中基于混合案例的分类推理模型

这篇文章提出了一种具有RUS和成本敏感的混合CBR模型，并在物联网环境中对成本敏感的反向传播神经网络通过特征加权进行交易。不平衡数据中的问题。使用六个真实数据集验证了所提议的模型。实验结果表明，所提议的模型更好而不是其他功能加权方法。KEywORdS人工神经网络（ANN），基于案例的推理（CBR），类不平衡，数据挖掘，物联网（IoT）