استفاده از روش هاي تركيب تصميم در بهبود طبقه بندي هاي سري زماني

چكيده انگليسي :

82 Using Decision Combination to Improve Time Series Classification Nazanin Asadi n asadi@ec iut ac ir Date of Submittion September 20 2012 Department of Electriacl and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language FarsiSupervisor Abdolreza Mirzaei mirzaei@cc iut ac irAbstract Outputs of real world processes can be considered as signals Describing these signals in the formof signal models or times series is a well known problem nowadays In these cases by time series we mean a sequence of data points measured typically at successive time instants spaced atuniform time intervals The importance of modeling time series has several reasons Among them these models are useful for simulating the source of signals and making a good theoretical basis forsignal processing systems The used model for these systems can be either deterministic model orstatistical one In this thesis for modeling training and classifying time series a statistical modelnamed Hidden Markov Model HMM is used Having a strong mathematical basis and its greatand varied applications especially in the field of voice recognition were some reasons for selectingHMMs On the other hand combining models and classifiers in order to improve the performanceand lower the error rate is used widely for a while in hard probelms In these methods severalweak classifiers are combined with each other to construct a strong good classifier If the basicclassifiers have enough difference and performance of each one is higher than a random model theobtained combined model expected to have higher performance in comparision to others Thoughmany researchers investigated classifier combinination in their papers combining classifiers withthe aim of obtaining the final classifier by means of internal structure of all basic classifiers whichare particularly used for time series is considered less yet Pervious methods wich use classifiercombination techniques do their combination in the final decision level or are not specialized fortime series structures Clustering is also a classification procedure in which the used data don thave any training labeled data and the purpose is to detect and classify similar data in the samegroup and the optimum case is when the internal distance among the data of the same clusters areminimum and the extra distance between different clusters is as much as possible Due to nothaving label for data in clustering this problem is more difficult than classification In this thesisthe first attempt was to obtain a good model for time series by using Hidden Markov Models whichcan best describe the given data and achieve a good performance and superiority compared to otherconsidered methods Second by using this training method and separating the models of differentclasses from each other a new technique for time series classification by the means of HiddenMarkov Models is proposed The third problem was clustering For this problem again by means ofHidden Markov Models and involving measures from pervious researches a new time seriesclustering method is proposed which take advantage of combining basic models and revealedbetter performance than other compared methods in the experiments Keywords Time series Hidden Markov Model classification training clustering

اسدي، نازنين

استفاده از روش هاي تركيب تصميم در بهبود طبقه بندي هاي سري زماني

مدل مخفي ماركوف , طبقه بندي , آموزش , خوشه بندي