資源描述：
《基于改進(jìn)神經(jīng)網(wǎng)絡(luò)的挖掘算法設(shè)計(jì)》由會員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在工程資料-天天文庫。
1、摘要：針對傳統(tǒng)的數(shù)據(jù)挖掘算法存在結(jié)構(gòu)復(fù)雜、耗時(shí)長、數(shù)據(jù)分析過程中易出現(xiàn)錯(cuò)誤，數(shù)據(jù)計(jì)算結(jié)果難以準(zhǔn)確表達(dá)結(jié)果等缺陷，提出結(jié)合神經(jīng)網(wǎng)絡(luò)在數(shù)據(jù)挖掘中的應(yīng)用方法。由于神經(jīng)網(wǎng)絡(luò)擁有對噪聲數(shù)據(jù)承受能力高、錯(cuò)誤率低等優(yōu)點(diǎn)，因此結(jié)合神經(jīng)網(wǎng)絡(luò)系統(tǒng)對數(shù)據(jù)挖掘算法進(jìn)行改進(jìn)設(shè)計(jì)可大幅度提高數(shù)據(jù)準(zhǔn)確性，該方法擁有結(jié)構(gòu)簡單、表述清晰、精準(zhǔn)度高等優(yōu)勢。為基于神經(jīng)網(wǎng)絡(luò)的數(shù)據(jù)挖掘算法的可行性進(jìn)行了嚴(yán)謹(jǐn)?shù)膶?shí)驗(yàn)分析，對實(shí)驗(yàn)數(shù)據(jù)進(jìn)行認(rèn)真的記錄和研究，實(shí)驗(yàn)結(jié)果表明，基于神經(jīng)網(wǎng)絡(luò)的挖掘算法相比傳統(tǒng)數(shù)據(jù)挖掘算法，其精度明顯提高，且整個(gè)過程耗時(shí)較短，由此可證實(shí)
2、基于神經(jīng)網(wǎng)絡(luò)的數(shù)據(jù)挖掘算法具有更高的實(shí)用性。關(guān)鍵詞：數(shù)據(jù)挖掘；神經(jīng)網(wǎng)絡(luò)；粗糙集；數(shù)據(jù)挖掘算法；數(shù)據(jù)計(jì)算；可行性分析中圖分類號：TN711734；TP183文獻(xiàn)標(biāo)識碼：A文章編號：10047373X(2018)1470143704DesignofminingalgorithmbasedonimprovedneuralnetworkHUANGWenfeng(HenanProvincialInstituteofScientific&TechnicalInformation,Zhengzhou450003,China)
3、Abstract：Inallusiontotheshortcomingsexistinginthetraditionaldataminingalgorithmforitscomplexstructure,longtimeconsumption,erroreasilyappearingindataanalysisprocess,anddifficultyinaccurateresultexpressionofdatacalculationresults,anapplicationmethodofcombining
4、neuralnetworkindataminingisputforward.Astheneuralnetworkhastheadvantagesofstrongabilitytowithstandnoisedata,andlowerrorrate,improvementdesignofthedataminingalgorithmbycombiningwiththeneuralnetworksystemcangreatlyimprovethedataaccuracy.Thismethodhastheadvanta
5、gesofsimplestructure,clearexpression,highprecisionandsoon.Arigorousexperimentanalysisforthefeasibilityofthedataminingalgorithmbasedonneuralnetworkisperformedonthebasisoftherecordedexperimentaldata.Theexperimentalresultsshowthattheminingalgorithmbasedonneural
6、networkhashigheraccuracyandshortertimeconsumptionthanthetraditionaldataminingalgorithm,whichcanconfirmthatthedataminingalgorithmbasedonneuralnetworkismorepractical.Keywords:datamining；neuraInetwork；roughset；dataminingalgorithm；datacalculation；feasibilityanal
7、ysisO數(shù)據(jù)挖掘算法是從數(shù)據(jù)庫的數(shù)據(jù)中提取隱含有用信息的過程?，F(xiàn)階段常用的數(shù)據(jù)挖掘算法和理論主要集中于粗糙集理論和遺傳算法等相結(jié)合的方式口]。為了解決傳統(tǒng)方法結(jié)構(gòu)復(fù)雜、耗時(shí)長、錯(cuò)誤率高等問題，提出改進(jìn)神經(jīng)網(wǎng)絡(luò)的數(shù)據(jù)挖掘算法。首先對神經(jīng)網(wǎng)絡(luò)進(jìn)行優(yōu)化和改進(jìn)，以便減少誤差，然后基于改進(jìn)后的神經(jīng)網(wǎng)絡(luò)設(shè)計(jì)數(shù)據(jù)挖掘算法。該算法擁有魯棒性強(qiáng)、精準(zhǔn)度高、抗噪能力強(qiáng)、數(shù)據(jù)承受能力高、錯(cuò)誤率低等優(yōu)勢，可廣泛應(yīng)用于機(jī)器學(xué)習(xí)等領(lǐng)域[2]o通過仿真實(shí)驗(yàn)對神經(jīng)網(wǎng)絡(luò)挖掘算法的功能進(jìn)行驗(yàn)證發(fā)現(xiàn)其對數(shù)據(jù)的搜索能力強(qiáng)大、擅長全局搜索，且對數(shù)據(jù)挖
8、掘的準(zhǔn)確性和實(shí)時(shí)性相對較高，可成功解決傳統(tǒng)方法中的諸多缺陷，具有較高的使用價(jià)值。1神經(jīng)網(wǎng)絡(luò)算法優(yōu)化設(shè)計(jì)結(jié)合神經(jīng)網(wǎng)絡(luò)自身特點(diǎn)以及遺傳算法的優(yōu)勢對神經(jīng)網(wǎng)絡(luò)進(jìn)行優(yōu)化[3]o本文默認(rèn)網(wǎng)絡(luò)結(jié)構(gòu)為拓?fù)浣Y(jié)構(gòu)，主要便于對隱層數(shù)及節(jié)點(diǎn)數(shù)進(jìn)行計(jì)算。由于人工干預(yù)常常會對網(wǎng)絡(luò)隱層數(shù)及節(jié)點(diǎn)數(shù)造成一定破壞，難以對其進(jìn)行準(zhǔn)確測量和計(jì)算，考慮通過遺傳算法對其進(jìn)行優(yōu)化從而避免人工干預(yù)次數(shù)[4]o首先，對初始化網(wǎng)絡(luò)結(jié)構(gòu)進(jìn)
