学海网 文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Clustering time series with hidden Markov models and dynamic time warping
免费下载此文档

Clustering time series with hidden Markov models and dynamic time warping

Given a source of time series data, such as the stock market or the monitors in an intensive care unit, there is often utility in determining whether there are qualitatively

Clsteuirg Timn Seeiesr wth HiiddenM akrvo oMels add Dnnaymc Timei Wapirn gTm Oiates, aLur Firoiu anda aPulR. Cheo CnomutpreS iecne Departmcetn,LGCRU iversntiy foMa sschauetts,sBox 34 610 mhArets,M 01A00-4631 0oftesa, loiu,rcohne@gc.umsas.sedu

Givne asuroec fo imte sreei sdta, ausc hsat h etsock amkrteor he tomntirso in n anienstiv ecaer uin, tteheris fotne uiltity in deteminirgn wethhr theee are qruailattveiy li dreen regitemsin t h daet aan di chnracaetriing thozes rgeims.e or Fexamlpe,on emghi ltik eo ktonw hewterht h evaiorus ndiiatcro ofs ap aitnet' sehlath emsuredao vre tmie aer ebig prondcudeby apa iten whtoi s liklye o tive ol roen htt asi ilelykto d i.e I tnis chse, ahtee is a prroirikno lwdge eo tfh eumner bo fergiem thast xiet sin thed at (awt),o na dth regimee t owihh cna ygveni tiem sriese elbons cganbe d eerminedt opt soh (cb ysimpy nlotig whethen rhe tpaitent livdeor ide)d .Hoewver, hest etowp ices of ienfromtaoi are not nlawya prssente.Co nsidera s stye mtahtp odurecsm utilvraatie re,alvaled utim eeriessby s elecitng noe f Koh idend arMko vmdoes l(HMMs, gener)tian dgta acaorcdin tog hattmo el,da d pnreodiicaly lrtnsaiiontnig ot a id reet HnM.M nlO yth etie mesiers rodpucde b yht syseem atr obeervsbae. lI panrtiulca,r Kndat e ihenditt oyft e hHMM gneeatirn gny aivegnti em sreesia re noto besrvblea. iGen v aes tfo ite smries peordcedu ybsu hc syatem,sth is paer ppresnes a tmthedo of artuoatimcllay etdemrniig Kn the,numer ofb gneearitn gHMsM,nda ofrl aerinngt h epraaetmesrof t osh HeMsM. n Aniitaliesti maeto fK i s otabned bi unsyuperisevdclus terignof teh tmi seeris usien gyndami ctie mwapirgn (TD)Wto assess siilmarty. In idadtioni ot porudinc ganes tmaiet o fK, tihs pocess yirles andi iniatlpar ttionini gof te dhat. Aa salte rsctieos wiln elpxalni, TDW i selrteadt HMo trMiainn galgrotihmsb u it wsaeek inr evsearl erpsecs. Ttehreofer,hte cultsrse bsed on aTD Wre liaekyl to octnai nimtases.k Tehseini iatlc ustles serrvea sipun tt o arpcoes tsha ttraisnone HMM o naehcc ulstr end ateritavelyi ovmes itm seereisbet ewn cluetesr sabes ondt hei lrikeiloodh gsivent hev riauo HMssM.U litamelty th preoces scnovereg tos an alcl uterisgn fothe data and geaernaitv medel (oht eHMM)s orfeac h o fth elustces.r Th remeainer dof teh ppea is rroanigzde as ofllwo. Sscetoin 2b ier yerveiwsH MM and sth elgaoitrhmsused t oidune HMMcsfr m oadt. aeScitno 3dsericbs eydnmac tiimewarp in agd ints su ase daistnaec eamusr ebtweee nmutilarvaie ttme iesrise fr tohep rposueof unsuepvirsedcl suetrng.i Scteion4 dsceribes naa logrthm thati seus

1Int roudtcon i DWT ot obtstrop ahetpr oecsso ftitg nHMMsto dat a continangi umltpleire giesm .ecSito 5 pnersens tte rhesuts lo fexeripment swiht hti spparaochu singa ti ricl data,a an decsiont6 c oncldue sad points tn fouurte wrko A d.scirtee hdied Manrko mvodl is de ene dy b seato stfaes atnda naphlaet bf ooutut psmbyols( aRibner1 989. Ea)c sthat esic hractearize db ytow rpobabilit yidtrsbitiuo

ns:te htarsinito dnsitrbutiion voer tsaet sad tnheem isiso nisdrtiubitn oovr ehteo tuuptsym obl.sA rando mso urec decsibrde yb usc h amdol geenreteasa seueqcn of oeuptt usmboyl assfo llwos:at ea hctim step teh seorcu ie isn no steta,ean da fer etimtitngan o tuup sytmolb ccoadirn tgo hete mssiino idsributtio noft e hcurrnet tsta, etehs orce uujpms t a neox sttatea ccrding ot otehtr anitisnod siribttuino foit scur ern ttsta.eSin ce th aceivtityof the sou creis bseorvdeindi rcelty,th rugoh te hseuqnce ef ooutup tymsbol, asn dht esqeeucn eof tstesa s noitdire ctylo bersavle, tbe htsaet sra eaisdto b hiddeen A.c ontiunuos MMHis di eern tfom ra idcrest eon ien tat hteh uotup tymsblso aeremi tet fdro ma robpabliity dnesty iisntadeo a dfistrbuitoni. e pWrfer teh eiscretd moeeld becausesthe arey simpel. Forrb to hiscdreet nd acnoinutousH MsMd namyic rpogarmmng aligrotihs exmstifor: cmouptin tge prhbobilaiyt o ofbesrvin gase qunce,e igev anm deol dnni tgehs atteseq unceet ht aamimizesxt hepro baiblti yfot h eivgne equesnc, wehe tnehmo dleis k own(thn eVteibira glroith) imnducin gtehH M Mhtt maaxiimzes (ocally)l te prohabilibytof te ghiev nesqunec (tee Bhaum-eWcl hlgoriath, mn expeacatitonmaxi-miztaio nlagriotm) The HhMMmo dl eednition ac nbe erdilyae xetnddeto t e mhulidimenstoina claes w,ere h avetorco f ymbsol si esittmd aetea hc tep, issneatdo f a insgle ysmbl.oT h assemupitn otaht alloswthi s imemdita extenseinoi s onditioncl anidepndeecne o vfairbleasg vie nht staete W.hile idhdn Mearokvmo elds ra eicherrr eperesntaitns ooft mei eries st(hey ddaa voca buarl ofys ttase, t)h ineudcitno agorlitm hash wtom jaro weaneksss:e

Hi2dedn MarokvM dols

第1页

免费下载Word文档免费下载:Clustering time series with hidden Markov models and dynamic time warping

(下载1-5页,共5页)

我要评论

相关文档

  • 语音识别技术研究

    series,SOwecanuse HMMhasadvantageofestablishingmodelfortimeHMMtoestablish models...[26]B.Hjuang.On t11ehiddenMarkovandDynamictimewarpingforspeech recognition.A...

  • 中英文翻译

    the need of many more training samples,but also long train time requirement...hidden Markov models (CDHMM),Dynamic time warping(DTW) and Biomimetic ...

  • 科学文献

    The parameters of these low level features evolve over time and form ...trajectories via Hidden Markov Models (HMM’s) or Dynamic Time Warping (...

  • 多权值神经元网络仿生模式识别方法在低训练样_省略_特...

    Hidden Markov Models ( HMMs) and Dynamic Time Warping (DTW) respectively. ...中国专利申请 using clustering techniques [ J ] . IEEE Trans. Acoustics , ...

  • 英文文献

    theory, such as Hidden Markov Model (HMM) and Dynamic Time Warping(DTW) ...Due to the number of ANN input nodes being fixed, therefore time ...

  • Improved speech recognition method for intelligen

    Up to now, most speech recognition is based on conventional linear system theory, such as Hidden Markov Model (HMM) and Dynamic Time Warping(DTW) . ...

  • 汉语孤立字语音识别技术的研究_图文

    speechrecognitionmethodsbasedandhiddenMarkov dynamictime warping(DTW)DTWalgorithm model(HMM)are deeplystudied.Thefast haslowcomplexityandisverysuitableforsmall-...

  • WarpNet Self-Organizing Time Warping

    minimize the the sum of distances over two utterances, or any time series...Dynamic Time Warping (DTW), or in systems based on Hidden Markov Models ...

  • Time Warping Under Dynamic Constraints

    each instant and then quantized over time using a hidden Markov model [8]...erent ways to yield di?erent time series. In time warping under dynamic ...

  • 基于声纹识别的Web身份认证系统设计

    【Key words】identity authentication; voiceprint recognition; Hidden Markov Model(HMM); Dynamic Time Warping(DTW) DOI: 10.3969/j.issn.1000-3428.2011.15...

站点地图 | 文档上传 | 侵权投诉 | 手机版
新浪认证  诚信网站  绿色网站  可信网站   非经营性网站备案
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.xuehai.net All Rights Reserved.  email
返回顶部