Dissertations / Theses on the topic 'Self-organizing map'
To see the other types of publications on this topic, follow the link: Self-organizing map.
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Self-organizing map.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
1
Farshad, Tabrizi Seyed Ramin. "The Probabilistic Supervised Self-Organizing Map, PSSOM." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1998. http://www.collectionscanada.ca/obj/s4/f2/dsk2/ftp01/MQ31828.pdf.
Full text
2
Choe, Yoonsuck. "Perceptual grouping in a self-organizing map of spiking neurons." Access restricted to users with UT Austin EID Full text (PDF) from UMI/Dissertation Abstracts International, 2001. http://wwwlib.umi.com/cr/utexas/fullcit?p3025202.
Full text
3
Schwardt, Martin. "Lösung ausgewählter Routenplanungsprobleme mit Hilfe der self-organizing map." [S.l.] : [s.n.], 2005. http://deposit.ddb.de/cgi-bin/dokserv?idn=975255126.
Full text
4
Pourkia, Javid. "A SELF-ORGANIZING MAP APPROACH FOR HOSPITAL DATA ANALYSIS." OpenSIUC, 2014. https://opensiuc.lib.siu.edu/theses/1553.
Full textAbstract:
In this work, we utilize Self Organized Maps (SOM) to cluster and classify hospital related data with large dimensions, provided by Medicare website. These data have published every year and it includes numerous measures for each hospital in the nationwide. It might be possible to unearth some correlations in health-care industry by being able to interpreting this dataset, for example by examining the relations between data of immunizations department to readmission records and hospital expenses. It is not feasible to make any sense from these measures altogether using traditional methods (2D or 3D charts, diagrams or graphs, different tables), because as a result of being human, we cannot comprehend more than 3 dimensions with naked eyes. Since it would be very useful if we could correlate the dimensions to each other to discover new patterns and knowledge, SOMs are a type of Artificial Neural Networks that can be trained using unsupervised learning to illustrate complex and high dimensional data by generating a low dimension representation of the training sample. This way, a powerful and easy-to-interpret visualization will be provided for healthcare officials to rapidly identify the correlation between different attributes of the dataset using clusters illustration
5
Žáček, Viktor. "Kohonenova samoorganizační mapa." Master's thesis, Vysoké učení technické v Brně. Fakulta elektrotechniky a komunikačních technologií, 2012. http://www.nusl.cz/ntk/nusl-219527.
Full textAbstract:
Work deal about self-organizing maps, especially about Kohonen self-organizing map. About creating of aplication, which realize creating and learning of self-organizing map. And about usage of self-organizing map for self-localization of robot.
6
Malondkar, Ameya Mohan. "Extending the Growing Hierarchical Self Organizing Maps for a Large Mixed-Attribute Dataset Using Spark MapReduce." Thesis, Université d'Ottawa / University of Ottawa, 2015. http://hdl.handle.net/10393/33385.
Full textAbstract:
In this thesis work, we propose a Map-Reduce variant of the Growing Hierarchical Self Organizing Map (GHSOM) called MR-GHSOM, which is capable of handling mixed attribute datasets of massive size. The Self Organizing Map (SOM) has proved to be a useful unsupervised data analysis algorithm. It projects a high dimensional data onto a lower dimensional grid of neurons. However, the SOM has some limitations owing to its static structure and the incapability to mirror the hierarchical relations in the data. The GHSOM overcomes these shortcomings of the SOM by providing a dynamic structure that adapts its shape according to the input data. It is capable of growing dynamically in terms of the size of the individual neuron layers to represent data at the desired granularity as well as in depth to model the hierarchical relations in the data.
However, the training of the GHSOM requires multiple passes over an input dataset. This makes it difficult to use the GHSOM for massive datasets. In this thesis work, we propose a Map-Reduce variant of the GHSOM called MR-GHSOM, which is capable of processing massive datasets. The MR-GHSOM is implemented using the Apache Spark cluster computing engine and leverages the popular Map-Reduce programming model. This enables us to exploit the usefulness and dynamic capabilities of the GHSOM even for a large dataset.
Moreover, the conventional GHSOM algorithm can handle datasets with numeric attributes only. This is owing to the fact that it relies heavily on the Euclidean space dissimilarity measures of the attribute vectors. The MR-GHSOM further extends the GHSOM to handle mixed attribute - numeric and categorical - datasets. It accomplishes this by adopting the distance hierarchy approach of managing mixed attribute datasets.
The proposed MR-GHSOM is thus capable of handling massive datasets containing mixed attributes. To demonstrate the effectiveness of the MR-GHSOM in terms of clustering of mixed attribute datasets, we present the results produced by the MR-GHSOM on some popular datasets. We further train our MR-GHSOM on a Census dataset containing mixed attributes and provide an analysis of the results.
7
Wang, Dali. "Adaptive Double Self-Organizing Map for Clustering Gene Expression Data." Fogler Library, University of Maine, 2003. http://www.library.umaine.edu/theses/pdf/WangD2003.pdf.
Full text
8
Mewes, Daniel, and Ch Jacobi. "Analyzing Arctic surface temperatures with Self Organizing-Maps: Influence of the maps size." Universität Leipzig, 2018. https://ul.qucosa.de/id/qucosa%3A31794.
Full textAbstract:
We use ERA-Interim reanalysis data of 2 meter temperature to perform a pattern analysis of the Arctic temperatures exploiting an artificial neural network called Self Organizing-Map (SOM). The SOM method is used as a cluster analysis tool where the number of clusters has to be specified by the user. The different sized SOMs are analyzed in terms of how the size changes the representation of specific features. The
results confirm that the larger the SOM is chosen the larger will be the root mean square error (RMSE) for the given SOM, which is followed by the fact that a larger number of patterns can reproduce more specific features for the temperature.Wir benutzten das künstliche neuronale Netzwerk Self Organizing-Map (SOM), um eine Musteranalyse von ERA-Interim Reanalysedaten durchzuführen. Es wurden SOMs mit verschiedener Musteranzahl verglichen. Die Ergebnisse zeigen, dass SOMs mit einer größeren Musteranzahl deutlich spezifischere Muster produzieren im Vergleich zu SOMs mit geringen Musteranzahlen. Dies zeigt sich unter anderem in der Betrachtung der mittleren quadratischen Abweichung (RMSE) der Muster zu den zugeordneten ERA Daten.
9
Bui, Michael. "Path finding on a spherical self-organizing map using distance transformations." Thesis, The University of Sydney, 2008. http://hdl.handle.net/2123/9290.
Full textAbstract:
Spatialization methods create visualizations that allow users to analyze high-dimensional data in an intuitive manner and facilitates the extraction of meaningful information. Just as geographic maps are simpli ed representations of geographic spaces, these visualizations are esssentially maps of abstract data spaces that are created through dimensionality reduction. While we are familiar with geographic maps for path planning/ nding applications, research into using maps of high-dimensional spaces for such purposes has been largely ignored. However, literature has shown that it is possible to use these maps to track temporal and state changes within a high-dimensional space. A popular dimensionality reduction method that produces a mapping for these purposes is the Self-Organizing Map. By using its topology preserving capabilities with a colour-based visualization method known as the U-Matrix, state transitions can be visualized as trajectories on the resulting mapping. Through these trajectories, one can gather information on the transition path between two points in the original high-dimensional state space. This raises the interesting question of whether or not the Self-Organizing Map can be used to discover the transition path between two points in an n-dimensional space. In this thesis, we use a spherically structured Self-Organizing Map called the Geodesic Self-Organizing Map for dimensionality reduction and the creation of a topological mapping that approximates the n-dimensional space. We rst present an intuitive method for a user to navigate the surface of the Geodesic SOM. A new application of the distance transformation algorithm is then proposed to compute the path between two points on the surface of the SOM, which corresponds to two points in the data space. Discussions will then follow on how this application could be improved using some form of surface shape analysis. The new approach presented in this thesis would then be evaluated by analyzing the results of using the Geodesic SOM for manifold embedding and by carrying out data analyses using carbon dioxide emissions data.
10
Tervonen, J. (Jaakko). "Exploring behaviour patterns with self-organizing map for personalised mental stress detection." Master's thesis, University of Oulu, 2019. http://jultika.oulu.fi/Record/nbnfioulu-201904131491.
Full textAbstract:
Abstract. Stress is an important health problem and the cause for many illnesses and working days lost. It is often measured with different questionnaires that capture only the current stress levels and may come in too late for early prevention. They are also prone to subjective inaccuracies since the feeling of stress, and the physiological response to it, have been found to be individual. Real-time stress detectors, trained on biosignals like heart rate variability, exist but majority of them employ supervised learning which requires collecting a large amount of labelled data from each system user. Commonly, they are tested in situations where the stress response is deliberately induced (e.g. laboratory). Thus they may not generalise to real-life conditions where more general behavioural data could be used.
In this study the issues with labelling and individuality are addressed by fitting unsupervised stress detection models at several personalisation levels. The method explored, the Self-Organizing Map, is combined with different clustering algorithms to find personal, semi-personal and general behaviour patterns that are converted to stress predictions. Laboratory biosignal-data are used for method validation. To provide an always-on type stress detection, real-life behavioural data consisting of biosignals and smartphone data are experimented on.
The results show that personalisation does improve the predictions. The best classification performance for the laboratory data was found with the fully personalised model (F1-score 0.89 vs. 0.45 with the general model) but for the real-life data there was no big difference between fully personal (F1-score 0.57) and general model as long as the behaviour patterns were mapped to stress individually (F1-score 0.60).
While the scores also validate the feasibility of SOM for mental stress detection, further research is needed to determine the most suitable and practical level of personalisation and an unambiguous mapping between behaviour patterns and stress.Tiivistelmä. Stressi on merkittävä terveysongelma ja syynä useisiin sairauksiin sekä työpoissaoloihin. Sitä mitataan usein erilaisilla kyselyillä, jotka kuvaavat vain hetkellistä stressitasoa ja joihin voidaan vastata liian myöhään ennaltaehkäisyn kannalta. Kyselyt ovat myös alttiita subjektiivisille epätarkkuuksille, koska stressintunteen, ja stressinaikaisten fysiologisten reaktioiden, on havaittu olevan yksilöllisiä. Reaaliaikaisia, biosignaalien kuten sykevälivaihtelun analyysiin perustuvia, stressintunnistimia on olemassa, mutta pääosin ne käyttävät ohjatun oppimisen menetelmiä, mikä vaatii jokaiselta järjestelmän käyttäjältä suuren stressintunteella merkityn aineiston. Stressintunnistimia myös usein testataan tilanteissa, joissa stressi on tahallisesti aiheutettua (esimerkiksi laboratoriossa). Siten ne eivät yleisty tosielämän tarpeisiin, jolloin voidaan käyttää yleisempää käyttäytymistä kuvaavaa aineistoa.
Tässä tutkimuksessa vastataan datan merkintäongelmaan sekä yksilöllisyyden huomioimiseen käyttäen ohjaamattoman oppimisen stressintunnistusmalleja eri yksilöimisen tasoilla. Käytetty menetelmä, itseorganisoituva kartta, yhdistetään eri ryhmittelyalgoritmeihin tavoitteena löytää henkilökohtaiset, osin henkilökohtaiset sekä yleiset käyttäytymismallit, jotka muunnetaan stressiennusteiksi. Menetelmän sopivuuden vahvistamiseksi käytetään laboratoriossa kerättyä biosignaalidataa. Menetelmää sovelletaan myös tosielämän stressintunnistukseen biosignaaleista ja älypuhelimen käyttödatasta koostuvalla käyttäytymisaineistolla.
Tulokset osoittavat, että yksilöiminen parantaa ennustetarkkuutta. Laboratorio-aineistolla paras luokittelutarkkuus löydettiin täysin yksilöllisellä mallilla (F1-pistemäärä 0.89, kun yleisellä 0.45). Tosielämän aineistolla täysin yksilöllisen (F1-pistemäärä 0.57) ja yleisen mallin, jossa käyttäytymismallien ja stressin välinen kuvaus määrättiin yksilöidysti (F1-pistemäärä 0.60), välinen ero ei ollut suuri.
Vaikka tulokset vahvistavatkin itseorganisoituvan kartan sopivuuden psyykkisen stressin tunnistamisessa, lisätutkimusta tarvitaan määräämään soveltuvin ja käytännöllisin yksilöimisen taso sekä yksikäsitteinen kuvaus käyttäytymismallien ja stressin välille.
11
Wandeto, John Mwangi. "Self-organizing map quantization error approach for detecting temporal variations in image sets." Thesis, Strasbourg, 2018. http://www.theses.fr/2018STRAD025/document.
Full textAbstract:
Une nouvelle approche du traitement de l'image, appelée SOM-QE, qui exploite quantization error (QE) des self-organizing maps (SOM) est proposée dans cette thèse. Les SOM produisent des représentations discrètes de faible dimension des données d'entrée de haute dimension. QE est déterminée à partir des résultats du processus d'apprentissage non supervisé du SOM et des données d'entrée. SOM-QE d'une série chronologique d'images peut être utilisé comme indicateur de changements dans la série chronologique. Pour configurer SOM, on détermine la taille de la carte, la distance du voisinage, le rythme d'apprentissage et le nombre d'itérations dans le processus d'apprentissage. La combinaison de ces paramètres, qui donne la valeur la plus faible de QE, est considérée comme le jeu de paramètres optimal et est utilisée pour transformer l'ensemble de données. C'est l'utilisation de l'assouplissement quantitatif. La nouveauté de la technique SOM-QE est quadruple : d'abord dans l'usage. SOM-QE utilise un SOM pour déterminer la QE de différentes images - typiquement, dans un ensemble de données de séries temporelles - contrairement à l'utilisation traditionnelle où différents SOMs sont appliqués sur un ensemble de données. Deuxièmement, la valeur SOM-QE est introduite pour mesurer l'uniformité de l'image. Troisièmement, la valeur SOM-QE devient une étiquette spéciale et unique pour l'image dans l'ensemble de données et quatrièmement, cette étiquette est utilisée pour suivre les changements qui se produisent dans les images suivantes de la même scène. Ainsi, SOM-QE fournit une mesure des variations à l'intérieur de l'image à une instance dans le temps, et lorsqu'il est comparé aux valeurs des images subséquentes de la même scène, il révèle une visualisation transitoire des changements dans la scène à l'étude. Dans cette recherche, l'approche a été appliquée à l'imagerie artificielle, médicale et géographique pour démontrer sa performance. Les scientifiques et les ingénieurs s'intéressent aux changements qui se produisent dans les scènes géographiques d'intérêt, comme la construction de nouveaux bâtiments dans une ville ou le recul des lésions dans les images médicales. La technique SOM-QE offre un nouveau moyen de détection automatique de la croissance dans les espaces urbains ou de la progression des maladies, fournissant des informations opportunes pour une planification ou un traitement approprié. Dans ce travail, il est démontré que SOM-QE peut capturer de très petits changements dans les images. Les résultats confirment également qu'il est rapide et moins coûteux de faire la distinction entre le contenu modifié et le contenu inchangé dans les grands ensembles de données d'images. La corrélation de Pearson a confirmé qu'il y avait des corrélations statistiquement significatives entre les valeurs SOM-QE et les données réelles de vérité de terrain. Sur le plan de l'évaluation, cette technique a donné de meilleurs résultats que les autres approches existantes. Ce travail est important car il introduit une nouvelle façon d'envisager la détection rapide et automatique des changements, même lorsqu'il s'agit de petits changements locaux dans les images. Il introduit également une nouvelle méthode de détermination de QE, et les données qu'il génère peuvent être utilisées pour prédire les changements dans un ensemble de données de séries chronologiquesA new approach for image processing, dubbed SOM-QE, that exploits the quantization error (QE) from self-organizing maps (SOM) is proposed in this thesis. SOM produce low-dimensional discrete representations of high-dimensional input data. QE is determined from the results of the unsupervised learning process of SOM and the input data. SOM-QE from a time-series of images can be used as an indicator of changes in the time series. To set-up SOM, a map size, the neighbourhood distance, the learning rate and the number of iterations in the learning process are determined. The combination of these parameters that gives the lowest value of QE, is taken to be the optimal parameter set and it is used to transform the dataset. This has been the use of QE. The novelty in SOM-QE technique is fourfold: first, in the usage. SOM-QE employs a SOM to determine QE for different images - typically, in a time series dataset - unlike the traditional usage where different SOMs are applied on one dataset. Secondly, the SOM-QE value is introduced as a measure of uniformity within the image. Thirdly, the SOM-QE value becomes a special, unique label for the image within the dataset and fourthly, this label is used to track changes that occur in subsequent images of the same scene. Thus, SOM-QE provides a measure of variations within the image at an instance in time, and when compared with the values from subsequent images of the same scene, it reveals a transient visualization of changes in the scene of study. In this research the approach was applied to artificial, medical and geographic imagery to demonstrate its performance. Changes that occur in geographic scenes of interest, such as new buildings being put up in a city or lesions receding in medical images are of interest to scientists and engineers. The SOM-QE technique provides a new way for automatic detection of growth in urban spaces or the progressions of diseases, giving timely information for appropriate planning or treatment. In this work, it is demonstrated that SOM-QE can capture very small changes in images. Results also confirm it to be fast and less computationally expensive in discriminating between changed and unchanged contents in large image datasets. Pearson's correlation confirmed that there was statistically significant correlations between SOM-QE values and the actual ground truth data. On evaluation, this technique performed better compared to other existing approaches. This work is important as it introduces a new way of looking at fast, automatic change detection even when dealing with small local changes within images. It also introduces a new method of determining QE, and the data it generates can be used to predict changes in a time series dataset
12
Wang, Jie. "Credit rating classification of China listed company with self-organizing map and discriminant analysis." Thesis, University of Macau, 2009. http://umaclib3.umac.mo/record=b2147584.
Full text
13
游鴻志. "Evolutionary Self-Organizing Map." Thesis, 1997. http://ndltd.ncl.edu.tw/handle/31927797953001438650.
Full textAbstract:
碩士中原大學
資訊工程學系
85
Since the concept of neural network was introduced, there are many neural network models developed and used broadly in different research areas. In 1973, Kohonen proposes a new generation of neural network models, called Self-Organizing Map (SOM). The SOM algorithm adds one kinds of fixed neighborhood relations among neurons into regular neural nets and learns new patterns under such neighborhood constraints. Although SOM has been proven to be effective in many applications, the fixedness of its neighborhood relations bring many inconveniences. This dissertation analyzes SOM from the viewpoint of graph theory and proposes a couple of dynamic SOM algorithms: “Evolutionary Self- Organizing Map algorithm ”and “Mate Self-Organizing Map algorithm”. In “Evolutionary Self-Organizing Map (EXOM) algorithm”, the weight functions of neurons and the neighborhood relations are combined as a neighborhood graph. As the training data are provided, this neighborhood graph can be dynamically evolved accordingly. Thus the initial state of neurons is not so critical and the final structure of this neural net will be adjusted to fit the training data automatically. Based on genetic algorithm, “Mate Self-Organizing Map (MSOM)”is proposed as another choice of SOM evolution strategies. MSOM mates two different neural nets according to their genes and breeds two children neural nets with different gene structures from their parents'''' . As genetic algorithm, such children neural nets of MSOM have chance to escape from local minimum (local-minimum -free property). From biology viewpoint, ESOM is one kind of asexual reproduction as well as MSOM is a sexual reproduction. With these two evolution strategies we can make SOM grow as biological population. Such flexible neural net structure not only patches the weakness of traditional SOM algorithm, but also makes possible some applications with uncertainty environment. For example, both evolution algorithms can be used in agent models to let agents with different knowledge or properties evolve and mate to produce their nest generations in an multiple agent architecture.
14
"Soft self-organizing map." Chinese University of Hong Kong, 1995. http://library.cuhk.edu.hk/record=b5888572.
Full textAbstract:
by John Pui-fai Sum.Thesis (M.Phil.)--Chinese University of Hong Kong, 1995.
Includes bibliographical references (leaves 99-104).
Chapter 1 --- Introduction --- p.1
Chapter 1.1 --- Motivation --- p.1
Chapter 1.2 --- Idea of SSOM --- p.3
Chapter 1.3 --- Other Approaches --- p.3
Chapter 1.4 --- Contribution of the Thesis --- p.4
Chapter 1.5 --- Outline of Thesis --- p.5
Chapter 2 --- Self-Organizing Map --- p.7
Chapter 2.1 --- Introduction --- p.7
Chapter 2.2 --- Algorithm of SOM --- p.8
Chapter 2.3 --- Illustrative Example --- p.10
Chapter 2.4 --- Property of SOM --- p.14
Chapter 2.4.1 --- Convergence property --- p.14
Chapter 2.4.2 --- Topological Order --- p.15
Chapter 2.4.3 --- Objective Function of SOM --- p.15
Chapter 2.5 --- Conclusion --- p.17
Chapter 3 --- Algorithms for Soft Self-Organizing Map --- p.18
Chapter 3.1 --- Competitive Learning and Soft Competitive Learning --- p.19
Chapter 3.2 --- How does SOM generate ordered map? --- p.21
Chapter 3.3 --- Algorithms of Soft SOM --- p.23
Chapter 3.4 --- Simulation Results --- p.25
Chapter 3.4.1 --- One dimensional map under uniform distribution --- p.25
Chapter 3.4.2 --- One dimensional map under Gaussian distribution --- p.27
Chapter 3.4.3 --- Two dimensional map in a unit square --- p.28
Chapter 3.5 --- Conclusion --- p.30
Chapter 4 --- Application to Uncover Vowel Relationship --- p.31
Chapter 4.1 --- Experiment Set Up --- p.32
Chapter 4.1.1 --- Network structure --- p.32
Chapter 4.1.2 --- Training procedure --- p.32
Chapter 4.1.3 --- Relationship Construction Scheme --- p.34
Chapter 4.2 --- Results --- p.34
Chapter 4.2.1 --- Hidden-unit labeling for SSOM2 --- p.34
Chapter 4.2.2 --- Hidden-unit labeling for SOM --- p.35
Chapter 4.3 --- Conclusion --- p.37
Chapter 5 --- Application to vowel data transmission --- p.42
Chapter 5.1 --- Introduction --- p.42
Chapter 5.2 --- Simulation --- p.45
Chapter 5.2.1 --- Setup --- p.45
Chapter 5.2.2 --- Noise model and demodulation scheme --- p.46
Chapter 5.2.3 --- Performance index --- p.46
Chapter 5.2.4 --- Control experiment: random coding scheme --- p.46
Chapter 5.3 --- Results --- p.47
Chapter 5.3.1 --- Null channel noise (σ = 0) --- p.47
Chapter 5.3.2 --- Small channel noise (0 ≤ σ ≤1) --- p.49
Chapter 5.3.3 --- Large channel noise (1 ≤σ ≤7) --- p.49
Chapter 5.3.4 --- Very large channel noise (σ > 7) --- p.49
Chapter 5.4 --- Conclusion --- p.50
Chapter 6 --- Convergence Analysis --- p.53
Chapter 6.1 --- Kushner and Clark Lemma --- p.53
Chapter 6.2 --- Condition for the Convergence of Jou's Algorithm --- p.54
Chapter 6.3 --- Alternative Proof on the Convergence of Competitive Learning --- p.56
Chapter 6.4 --- Convergence of Soft SOM --- p.58
Chapter 6.5 --- Convergence of SOM --- p.60
Chapter 7 --- Conclusion --- p.61
Chapter 7.1 --- Limitations of SSOM --- p.62
Chapter 7.2 --- Further Research --- p.63
Chapter A --- Proof of Corollary1 --- p.65
Chapter A.l --- Mean Average Update --- p.66
Chapter A.2 --- Case 1: Uniform Distribution --- p.68
Chapter A.3 --- Case 2: Logconcave Distribution --- p.70
Chapter A.4 --- Case 3: Loglinear Distribution --- p.72
Chapter B --- Different Senses of neighborhood --- p.79
Chapter B.l --- Static neighborhood: Kohonen's sense --- p.79
Chapter B.2 --- Dynamic neighborhood --- p.80
Chapter B.2.1 --- Mou-Yeung Definition --- p.80
Chapter B.2.2 --- Martinetz et al. Definition --- p.81
Chapter B.2.3 --- Tsao-Bezdek-Pal Definition --- p.81
Chapter B.3 --- Example --- p.82
Chapter B.4 --- Discussion --- p.84
Chapter C --- Supplementary to Chapter4 --- p.86
Chapter D --- Quadrature Amplitude Modulation --- p.92
Chapter D.l --- Amplitude Modulation --- p.92
Chapter D.2 --- QAM --- p.93
Bibliography --- p.99
15
Cheng, Wei-Chen, and 鄭為正. "Distance Invariant Self-organizing Map." Thesis, 2010. http://ndltd.ncl.edu.tw/handle/33410109389825158018.
Full textAbstract:
博士臺灣大學
資訊工程學研究所
98
This dissertation presents a distance invariant manifold that preserves neighboring relationships among data patterns. Since all input patterns have their corresponding cells in the manifold space, the neighboring cells of the input pattern resembles that of the output patterns. The manifold is invariant under the translation, rotation and scale of the pattern coordinates. And the neighboring relationships among cells are adjusted and improved in each iteration according to the algorithm of reduction of the distance preservation energy. This dissertation also extends the algorithm to presents a MLP kernel. It maps all patterns in a one class into a single point in the output layer space and maps different classes into different points. These widely separated class points can be used for further classifications. The kernel is a layered feed-forward network. Each layer is trained using class differences and is trained independently layer after layer using a bottom-up construction. The value of class labels are not used in the training process. Therefore, this kernel can be used in separating multiple classes.
16
Hui-Ling, HSU, and 許惠玲. "Semantic Indexing using Self-Organizing Map." Thesis, 2001. http://ndltd.ncl.edu.tw/handle/24872588529247968441.
Full textAbstract:
碩士國立臺灣大學
資訊工程學研究所
89
The information in literary works is rich. The task to read between the lines is challenging even for the most sophisticated system such as human brains. The main idea of this paper is to illustrate the design of a corpus-based method to find semantic structures in the complete works of Mark Twain (Samuel L. Clemens, 1835-1910), who was famous for his extraordinary sense of humor as well as social concern. Self-organizing Map (SOM) is applied to investigate the relations of vocabulary within the context of Mark Twain’s works. SOM make words cluster according to their context vectors, and are encoded based on their unique contextual structures. We call this practice "semantic encoding". Based on semantic encoding, a system for "semantic search" is further designed to search for passages in Mark Twain’s corpus which are closely related to user’s input. The passages found may have nothing in common with the input wording, however, they share similar semantic meaning.
17
Chen, De-Hua, and 陳德華. "Self-Organizing Map Networks for Symbolic Data." Thesis, 2009. http://ndltd.ncl.edu.tw/handle/71732493029297668562.
Full textAbstract:
博士中原大學
應用數學研究所
97
Abstract The Kohonen’s self-organizing map (SOM) is a competitive learning neural network that uses a neighborhood lateral interaction function to discover the topological structure hidden in the data set. It is an unsupervised approach. In general, the SOM neural network is constructed as a learning algorithm for numeric (vector) data. However, except these numeric data, there are many other data types such as symbolic data. The SOM algorithm cannot treat the symbolic data. In this dissertation we are interested in considering a modified SOM for symbolic data. Thus, a new SOM algorithm, called a symbolic SOM (S-SOM), is proposed to deal with symbolic data. El-Sonbaty and Ismail [16] proposed a concept of cluster center that is a structure where the cluster center contains events and associated memberships. We will use the cluster center structure as a symbolic neuron. We can use different associated memberships to display different symbolic neurons. The distance measures from Gowda and Diday’s dissimilarity [6] and Yang et al. [9] are used. Lateral interaction is an important factor of learning. It is excited by the input data and neurons. Fan et al. [5] proposed the suppressed fuzzy c-means (S-FCM) which expand the largest membership degree and suppress the others. We used the suppressed idea to create a learning rule of neurons. That is, to expand the largest associated membership and to suppress the others. Thus, we use the suppressed concept of associated membership to create a learning rule of symbolic neuron. This is the main spirit of learning rule for symbolic neurons. Finally, we apply the symbolic SOM to real examples. The results show feasibility of our symbolic SOM in real applications.
18
Lo, Yung-Ho, and 羅永和. "Chinese Document Clustering Using Self-Organizing Map." Thesis, 2004. http://ndltd.ncl.edu.tw/handle/97883278971622717173.
Full textAbstract:
碩士國立高雄第一科技大學
資訊管理所
92
The 21st centenary is an age of information explosion. The continuous growth in the size and use of the Internet is creating difficulties in the search for information. Currently, the problem which the users encountered, are not lack of information but too much information. There is a need for automatic procedures that allow users to retrieve the information from the rich information sources. An effective algorithm to organize the structure of information and assist user to search information is therefore particularly important. As well known Category map developed based on Kohonen’s self-organizing map (SOM) has been proven to be a promising browsing tool for information retrieval. The SOM algorithm automatically compresses and transforms a complex information space into a two-dimensional graphical representation. Such graphical representation provides a user-friendly interface for users to explore information repository. In this study, we applied SOM artificial neural networks algorithm to organize the Chinese plant documents into a two-dimension display map, as a visual tool to assist user to fulfill their information need. In the preprocessing stage, we focused our work on Chinese word segmentation and removed the stopwords for constructing a specific plant domain corpus, and vectored the documents with this corpus as the input value for SOM artificial neural networks. The result of this study show that Chinese plant document map has good recall rates and precision rates in our experiment.
19
Li, Dong-Lin, and 李東霖. "Adaptive Self-Organizing Map and Its Applications." Thesis, 2006. http://ndltd.ncl.edu.tw/handle/25233280500026750414.
Full textAbstract:
碩士國立中興大學
電機工程學系所
94
Self-organizing neural network is one of the methods frequently used in data clustering. In this thesis, we present a new method to improve the self-organizing map algorithm. Instead of the 2-D neighborhood topology in the conventional self-organizing map, a 3-D 6-neighbor topology is adopted in our approach. To avoid the dead (non-functional) neurons and to represent the training data more effectively, the number of neurons and the links between the neurons will be adjusted automatically during the process of the competitive learning by using a self-constructing model.
20
Ke, Kuo-Lung, and 柯國隆. "Research on Topic Oriented Self-organizing Map." Thesis, 2010. http://ndltd.ncl.edu.tw/handle/36641514148320184803.
Full textAbstract:
碩士國立高雄大學
資訊管理學系碩士班
98
Text document clustering is a basic operation of text processing and is widely applied in data visualization, theme identification, text summarization, hierarchy generation, etc. However, it will be inconvenient for users to find a document after clustering without proper labeling of topics. Moreover, there exist hierarchical relationships between document clusters. In this work, we will propose an adaptive self-organizing map model, namely the topic-oriented self-organizing map (TOSOM), that can adaptively expand the map laterally and hierarchically according to the topics of clusters, rather than the data distributions used in traditional adaptive self-organizing maps such as growing hierarchical self-organizing map (GHSOM). We conducted experiments on the Reuters-21578 dataset and obtained promising result.
21
Lin, Chung-Fu, and 林長富. "Fractal Image Compression using Self-Organizing Map." Thesis, 2006. http://ndltd.ncl.edu.tw/handle/50071198199820242247.
Full textAbstract:
碩士義守大學
資訊工程學系碩士班
94
Fractal Image Compression possesses the advantages of high compression ratio, low loss ratio, and fast decompression process. The use of exhaustive search in the encoding process results in a long encoding time. However, the encoding time can be reduced with a limiting searching space. Searching space can be limited to a certain cluster through clustering the domain pool to effectively shorten the comparison time and achieve the goal of reducing encoding time. In this paper, the clustering of domain pool is implemented by Self-Organizing Map. Each domain block is transformed to frequency domain, and a small amount of AC coefficients is used to train the SOM to construct a frequency-based SOM, and then transforms the frequency-based SOM to construct a new spatial-based one. The frequency-based SOM can cluster domain blocks according to edge direction, and the spatial-based one retains most of the feature. The SOM not only accelerates the encoding speed effectively, but also preserves the image quality. And the spatial-based clustering method saves the time on transforming image blocks to frequency domain during encoding process.
22
Hsu, Hsuanming, and 許軒銘. "Vector Quantization Based On Self-Organizing Map." Thesis, 2012. http://ndltd.ncl.edu.tw/handle/87584128893710536844.
Full textAbstract:
碩士義守大學
資訊工程學系
100
Vector quantization has the advantages of high compression ratio and fast decompression, but in a codebook generation and encoding process must be a global search, making the coding process lengthy. If we can effectively reduce the search range of the codebook generation and encoding will be able to reduce the time spent by the codebook generation and encoding. On the other hand, the traditional vector quantization to generate codebook method that is using the LBG algorithm. The initialization of codebook is very important. What if the initialization is selected to focus too much on the codebook training they would not have adequate representation. The characteristics of self-organizing map that is between neurons of the winning neuron and surrounding area will have similar characteristics, as long as sufficient number of training, as well as to update the weightings properly, and we can have a good codebook. This study intend to propose the use of the codebook based on self-organizing map, by means of different update of weightings, combined with vector quantization, and then improve the codebook generation and maintain image quality.
23
Liao, Wen-Chung, and 廖文忠. "Extended Self-Organizing Map for Transactional Data." Thesis, 2012. http://ndltd.ncl.edu.tw/handle/95662897650762158331.
Full textAbstract:
博士國立雲林科技大學
管理研究所博士班
100
In many application domains, transactions are the records of personal activities. Transactions always reveal personal behavior customs, so clustering the transactional data can divide individuals into different segments. Transactional data are often accompanied with a concept hierarchy, which defines the relevancy among all of the possible items in transactional data. However, most of clustering methods for transactional data ignore the existing of the concept hierarchy. Owing to the lack of the relevancy provided by the concept hierarchy, clustering algorithms tend to separate some similar patterns into different clusters. Besides, their clustering results are not easy to be viewed by users. The purpose of this study is to propose an extended SOM model which can handle transactional data accompanied with a concept hierarchy. The new SOM model is named as SetSOM. It can project the transactional data into a two-dimensional map; in the meanwhile, the topological order of the transactional data can be preserved and visualized in the 2-D map. Besides transactional data, we apply the SetSOM to categorical data and mixed data. Experiments on synthetic and real world datasets were conducted, and the results demonstrated the SetSOM outperforms other SOM models and some state-of-art algorithms in execution time, visualization, mapping, and clustering.
24
Kuo, Kuan-hui, and 郭冠輝. "Application of Using Self-Organizing Feature Map in Concept Map." Thesis, 2004. http://ndltd.ncl.edu.tw/handle/68288164747667026745.
Full textAbstract:
碩士國立臺南大學
資訊教育研究所教學碩士班
92
To achieve the learning objectives, tests are often used to measure the achievements of the learners in traditional way of teaching and learning. The result of testing not only provides the information about what learners have achieved but also identify learners’ weaknesses. A good test puts emphasis on exploring learners’ learning process in order to go into the cognitive behavior of the learners and to evaluate what learners have achieved. The Self-Organizing Feature Map of neural network is applied to integrating professionals’ and teachers’ different points of view on the concept relation. Different professionals and teachers have dissenting views on concept relation. Reasonable and objective concept relation is to be found by means of concept clustering of Self-Organizing Feature Map of neural network to put an end to the interfering factor. The program develops the best remedy instructions rout suggestion by using topology map of Self–Organizing Feature Map. The effective learning path is provided to learners as suggestions.
25
Wang, Sheng-Hsuan, and 王勝玄. "Clustering of Self-Organizing Map on Mixed Data." Thesis, 2005. http://ndltd.ncl.edu.tw/handle/95452369784589056271.
Full textAbstract:
碩士國立雲林科技大學
資訊管理系碩士班
93
The visualization-induced SOM (ViSOM) is a non-linear multi-dimensional projection method, extended from self-organizing map (SOM). It overcomes the drawbacks that the structure of the clusters may not be apparent and the nodes often spread around the 2-D map in the SOM. The objective of the ViSOM is to preserve the data structure as well as the topology as faithfully as possible. Even so, it still cannot express reasonably the distance or similarity of categorical data and preserve the structure of categorical data. In this study, the extended ViSOM is proposed to overcome these shortcomings. We utilize the concept hierarchies to define and calculate the distance of categorical values and preserve the structure of mixed data as well as the topology of trained EViSOM map as faithfully as possible. In addition, we perform clustering based on the output map generated by the network and evaluate the clustering result. Experimental results on two synthetic and two real datasets demonstrate that the proposed clustering algorithm is able to cluster mixed data better than the traditional SOM and ViSOM do. In addition, our algorithm better reveals the cluster structure and the clustering quality than traditional approaches, with respect to manual or automatic clustering of the trained map.
26
Chen, De-Hua, and 陳德華. "Self-Organizing Map Networks for Mixed Feature Data." Thesis, 2003. http://ndltd.ncl.edu.tw/handle/4psqvb.
Full textAbstract:
碩士中原大學
應用數學研究所
91
In this paper we propose a Kohonen’s self-organizing map (SOM) for mixed feature (symbolic type and fuzzy type) data. The distance measures proposed in Gowda ＆ Diday[3,4], El-Sonbaty ＆ Ismail[2], Yang, Hwang ＆ Chen[9] and Yang ＆ Ko[10] are used in this paper. Based on these distance measuces we proposed the mixed feature data SOM (MFD-SOM). We then propose a modified type of MFD-SOM method, called the modified self-organizing map for mixed feature data (MFD-M-SOM). On the other hand, we use fuzzy-soft learning method of Wu ＆ Yang[8], and then create the fuzzy-soft self-organizing feature map for mixed feature data (MFD-FS-SOM). On the learning method of neuron, we propose a new transformation method that adjust the neurons. Numerical examples and comparisons are given. These algorithms are also applied to real data with mixed feature data type. Finally, numerical results show that MFD-M-SOM has better performance than MFD-FS-SOM and MFD-FS-SOM is better than MFD-SOM.
27
Kaun, Wen-yu, and 關雯尤. "Object-Based Image Retrieval Using Self-Organizing Map." Thesis, 2012. http://ndltd.ncl.edu.tw/handle/05698743129249762603.
Full textAbstract:
碩士逢甲大學
資訊電機工程碩士在職專班
100
Content-based images retrieval (CBIR) has been widely used in many application fields. Yet, in commercial photography, ornaments and décor are often used to better the vision of the product as a whole. The images of the digital archive system for Taiwan flower anthography group are usually accompanied by other background objects that have nothing to do with the target plant. The background noise will decrease the precision rate of image retrieval. Therefore, we propose a method based on Visual Attention Model to extract image area of interest as training datasets, thus improving the retrieval precision rate. The number of digital images is growing rapidly in social networks due to the popularity of the Internet and digital cameras. To deal with the large amount of image data and computation cost, we choose the Self-Organizing Map (SOM) as our unsupervised learning algorithm. As an efficient Artificial Neural Network approach, SOM is useful for visualizing low-dimensional views of high-dimensional data. Our experiments show good performance results.
28
Hsiao, Ya Wen, and 蕭雅文. "The construction of knowledge map in medical information with self-organizing map." Thesis, 2008. http://ndltd.ncl.edu.tw/handle/94723697246861377148.
Full textAbstract:
碩士長庚大學
企業管理研究所
97
The World Wide Web contains a large number of documents that are dynamic, unregulated nature and rapid proliferation. It is increasingly difficult to search for relevant medical information. Many tools have been developed to help users search for useful information, but most of them are still not efficient. The goal of this paper is to describe an architecture designed to integrate text mining, an automatic thesaurus, and Growing Hierarchical Self-Organizing Map (GHSOM) technologies to provide searchers with fine grained results. Thus, we present a content-base and easy-to-use map hierarchy for medical documents from http://mag.udn.com/mag/life/. Meanwhile, an enhanced topic selection module and a web-based user interface are also proposed. User can browse a graphical display of medical-related topics. Besides, the construction of knowledge representation has to be automatic in order to efficiently handle the fast-growing and changeable medical information. Lastly, we apply a task experiment to evaluate the medical maps. The result shows that the medical map approach outperforms the keyword search approach.
29
Tew, Chee-Yuen, and 趙志運. "A Self-Organizing Feature-Map-Based Neuro-Fuzzy System." Thesis, 2001. http://ndltd.ncl.edu.tw/handle/85585354995827620877.
Full textAbstract:
碩士淡江大學
電機工程學系
89
One of the challenges that arise in designing a fuzzy system is the trade-off between computational efficiency and performance. Basically, the more rules, the more powerful the fuzzy system becomes. However, the price paid for the high performance is that the computational load becomes extremely large. In this thesis, the author proposes an appealing and easy solution to solve the dilemma. This thesis presents an efficient scheme for fuzzy modeling by using the Kohonen’s self-organizing feature map (SOM) algorithm through its vector quantization feature and its topological property. The vector quantization feature of feature maps is used to search a good supply of most representative cluster centers. Then the topology-preserving feature is fully utilized to select a set of most influential rules to be used in the computation of system outputs. By behaving this way, the proposed SOM-based fuzzy system provides an appealing solution to the trade-off between computational efficiency and performance. Besides, in order to further accelerate the computation of system outputs, the author also proposes a fast winner finding method to quickly locate winners in a feature map. To demonstrate the effectiveness of the proposed SOM-based fuzzy system, the proposed SOM-based fuzzy systems is applied on the problems of pattern recognition and system identification. From the simulation results show that the proposed SOM-based fuzzy system outperforms the conventional one-dimensional structured neuro-fuzzy systems on the recognition rates and the learning speed.
30
Wu, Po-hung, and 吳柏宏. "Self-Organizing Map on Auditory-Scene based Sound Segregation." Thesis, 2008. http://ndltd.ncl.edu.tw/handle/06644794740660182920.
Full textAbstract:
碩士國立交通大學
電信工程系所
97
During the past decade, detailed characteristics of auditory perception have been largely incorporated into speech processing algorithms to enhance their performance. For example, in the field of sound segregation, algorithms good for the condition of multiple microphones, such as independent component analysis (ICA), are often used and show satisfactory performance. However, the truth is human has no problems in segregating mixed sounds with only one ear. In this thesis, we design such a monaural speech segregation system based on an auditory perceptual model. Various spectral-temporal cues extracted from the model are used for monaural speech segregation. Then, a self-organizing feature map neural network is utilized to mimic the neural function in segregating and clustering a mixed sound into separated sounds. At the end, we demonstrate our system’s performance by comparing the separated sound with original sound.
31
Wu, Chih-Ting, and 吳智婷. "Using Self-organizing Map to Diagnose Abnormal Engineering Change." Thesis, 2013. http://ndltd.ncl.edu.tw/handle/uggtke.
Full textAbstract:
碩士德明財經科技大學
資訊管理系
102
Engineering change can’t be avoided in the product life cycle. It will impact the enterprise profitability. Using Self-organizing Map and engineering change history information was established one model. It will diagnose and monitor engineering change for the enterprise. The product life cycle is 3-6 months in the case. The four months for historical data per group in the experiment, one was selected from January-April (2343 records) of control group, others were selected from May to August (3642 records) and from September to December (3864 records) for the experimental two group. We established self-organizing map model (10 X 10 cells) by the SOM toolbox Matlab environment. Data recognition success rate were 96% and 95% for the experimental groups. Because some historical data didn’t exist so data is unrecognized. It has excellent capability, using self-organizing map to diagnose abnormal engineering. We analyzed abnormal engineering change by the decision tree in the case. Mainly causes are specific customer and specific change items. Therefore, it is important to monitor that engineering change state in the enterprise.
32
Kao, Huey-Shan, and 高慧珊. "Estimation of Evaporation using a Self-Organizing Map Network." Thesis, 2007. http://ndltd.ncl.edu.tw/handle/68323293614423494909.
Full textAbstract:
碩士國立臺灣大學
生物環境系統工程學研究所
95
The phenomenon of evaporation is an important factor that affects the distribution of water in hydrological cycle and plays a key role in agriculture and water resource management. The tranditional evaporation formulas usally neglect the non-linear characteristics in the nature. In this study we propose the self-organizing map(SOM) network to estimate daily evaporation. First, the daily meteorological data from climate gauges were collected as inputs of the SOM and then classified into topology map based on their similarities to investigate their potential property. To effectively and accurately estimate the daily evaporation, the connected weights between the cluster in topology layer with output layer were trained by using the linear regression method. In addition, we bulit enforced Self-Organizing Map (ESOM) to strength mapping spaces for these extremely data and compared with Modified Penman (FAO,1984) and Penman-Monteith (ICID,1994). The results demonstrated that the topology structures of SOM and ESOM could give a meaningful map to present the clusters of meteorological variables and the networks could well estimate the daily evaporation based on the input meteorological variables used in this study. In comparing the performances of these four models, the ESOM provides the best performance (RMSE=1.15mm/day，MAE=0.87 mm/day). The ESOM performance is also well in estimating long term evaporation. We have the suitability of using these models in other areas where their evaporations are different widely from the original station, the estimation, however, are not well as the one we use in the built station. This result suggests that the network must be adequately trained before it is used to estimate the local evaporation.
33
Chen, Jiun-Hung, and 陳俊宏. "Lips detection using self-organizing map and reinforcement learning." Thesis, 1999. http://ndltd.ncl.edu.tw/handle/28663252445109968468.
Full textAbstract:
碩士國立臺灣大學
資訊工程學研究所
87
Traditional lips detection methods consist of a binary classifiers which can classify lips and nonlips followed by some search algorithms. The search algorithms may depend on some face anatomatical information or heuristics. They do not learn any structural inforamiton while searching. A new lips detection approach is proposed. It solves classificaition and search at the same time based on structural infomation in faces. There are three main parts in this proposed approach. First, face detection is based on color and shape information and autocorrelation functions and golbal maximum difference are used as features for face image. Second, features are then clustered by self organizing map to form states. Third, by modeling the lips detection as a Markovian decision process, reinforcment learning is applied to find a optimal policy. The capability of this approach is demostrated by experimental results.
34
Chen, Wei-Yi, and 陳崴逸. "Application of Self-Organizing Map to Option Implied Volatility." Thesis, 2007. http://ndltd.ncl.edu.tw/handle/03620076200903578968.
Full textAbstract:
碩士國立交通大學
資訊管理研究所
95
In this study we investigate the lead-lag relations between the index option market and the stock market at the aggregate level. We could forecast the fluctuation of Taiwan Stock Market Index if the relations did exist. We apply the diagram constructed by volatility, the combination of the implied volatility of call and put, to represent the option market’s view for future stock market movements and discover their relations. Investors would long call when they expect the future price of spot market to soar. Thus, the implied volatility of call would rising. If the implied volatility of call is higher than put, the investors will long call when they expect the future price of spot market to decline. Thus, the implied volatility of put would rising, and the implied volatility of put is higher than call. Therefore, we could use self-organizing map to classify diagram constructed by implied volatility, checking the trading signal for next day and producing real trading suggestion. Comparing our strategy with buy-and-hold strategy, our strategy is better .
35
Tseng, Shih-Yu, and 曾士育. "Knowledge Discovery with Self-Organizing Map in Investment Strategy." Thesis, 2003. http://ndltd.ncl.edu.tw/handle/81886226686155662648.
Full textAbstract:
碩士國立高雄第一科技大學
資訊管理所
91
Financial investment is a knowledge-intensive industry. In the past years, with the electronic transaction technology advances, vast amount of transaction data have been collected and the emergence of knowledge discovery technology sheds light toward building up a financial investment decision support system. Data of financial markets are essentially time-series which bring more challenges than the traditional discrete data for uncovering the hidden knowledge. In this research, Taifex Index in Taiwan Futures Exchange K-chart patterns as the target dataset, we tackle with these challenges by proposing an integrated solution on the basis of knowledge-discovery methodology which supports four important tasks of data mining: clustering, classification, forecasting and visualization. In order to provide a decision maker the functionality of visualization, this project utilizes self-organization map that can transform high- dimensional, complicated, and nonlinear data onto low-dimensional ones with topology preservation. For clustering, the silhouette coefficient algorithm are applied to validate the clusters. Following, the trading signals are classified by performing pattern-match with K-Chart patterns in the trained SOM and the sliding-window data. Finally, the closing price in the next day is predicted based on the first 24 days pattern. In contrast to related work, we not only endeavor to improve the accuracy of classification of trading signals, we also are in an attempt to maximize the profits of trading. The resulting intelligence investment decision support system can help fund managers and investment decision-makers of national stable funds make the profitable decision. In addition, financial experts can benefit from the ability of verifying or refining their tacit investment knowledge offered by the uncovered knowledge.
36
Liao, Jyun Jie, and 廖俊傑. "Application of Dynamic Self-Organizing Map in Skeleton Extraction." Thesis, 2006. http://ndltd.ncl.edu.tw/handle/56992927686132474577.
Full textAbstract:
碩士南台科技大學
資訊工程系
94
Skeleton shape extraction technique was widely adopted in many application such as object modeling、character recognition、machine vision and computer animation .The thinning process always be used in skeleton extraction, but it often distorts the local shapes of an observed pattern. This is an inherent defect for all thinning algorithms. Therefore, a Dynamic Self-Organizing Map(DSOM) was proposed to extract skeleton precisely. But the process speed of DSOM is too slow. In order to speed up the process speed, the methods of feature-points estimation and area segmentation were proposed in this paper. Feature-points estimation process estimates the position of output neurons instead of random decision. The area segmentation process avoids the fault when DSOM used in crossing or loop pattern. In the experimental results, the tools、alphanumeric and Chinese characters are used as test patterns. The results show that the proposed methods improve the process speed of DSOM and it wouldn’t distort the local shapes of an observed pattern.
37
Hsieh, Ming-Hsun, and 謝明勳. "Using Hierarchical Modified Self-Organizing Map in Skeleton Extraction." Thesis, 2004. http://ndltd.ncl.edu.tw/handle/12725168864478341551.
Full textAbstract:
碩士南台科技大學
電子工程系
92
骨格抽出の技術はたくさん応用領域に広く採用されています。たどえばオブジェクト指向モデリング技術 (object modeling)、文字識別(character recognition)及びコンピュータアニメショーン(computer animation)などいろいろな応用領域に採用されています。しかし、伝統のthinningで抽出された骨格はいつも要らないの支線を分岐し、交差点の場合も変形を出ってくる事がありますから、特徴の獲得は不安定になります。だから、本論文では、「階層式修正型自己組織化マップ」を提出してこのようなの問題を解決する。 「階層式修正型自己組織化マップ(HMSOM)」はSOMとMSOMの二つの層から構成する。それで、第１、２層のネットワークとコホネンのSOMは同じ構造であり、二つの入力層と出力層から構成する。第１層では、SOMによってパターンの画素を入力データに。そして、前処理でニューロンのN個の数を概算し、出力ニューロンと画素の位置関係を考慮して、競合学習の基礎から、パターンをサブパターンにN個分割する。第２層では、１つづづのサブパターンをMSOMネットワークに入力させるし、ニューロンの数を１つにする。最後は、MSOMの可増加ニューロン数の特性によって、第１層の出力ニューロン数の不足を訂正する。それは、ニューロン数の不足は骨格の不完全させるの原因の１つ。しかし、ここで分枝がない骨格を抽出するだけ。だから、後処理で隣接のニューロンは互いに繋がっているの骨格は分枝が出って来る。しかし、交差点の所は変形のノードになるという問題が現れる。これは個別組合性の最適な問題が出って来る。だから、このような問題のために、変形のノードが新しいのノードに代わって。その問題を改善します。 本論文では工具、漢字、数字とか、英語文字などの画像を実験の対象にし、ノイズの影響、パターンを撮るの環境の光の不足と印刷の品質の劣化などの情況も考えている。実験の結果からを見ると、本論文の方法は確かに伝統のthinningより良いです。それで、SOMがノイズの影響を対抗が出来るのもはきり分かりました。しかし、システムの処理時間は長い時間がかかりますので、将来はこの問題を解決しる目標を目指します。
38
Su, Shi-Yong, and 蘇軾詠. "Data Visualization using Swarm Intelligence and the Self-Organizing Map." Thesis, 2004. http://ndltd.ncl.edu.tw/handle/19226849760754999861.
Full textAbstract:
碩士國立中央大學
資訊工程研究所
92
Social insects (or animals) provide us with a powerful concept to create decentralized systems of simple interacting, and often mobile, agents (e.g. ants, birds, etc.) The study of their behaviors provides us with effective tools for solving many difficult problems such as optimization, etc. More and more researchers are interested in this exciting way of achieving a form of swarm intelligence (i.e. the emergent collective intelligence of groups of simple agents.) They have created computer simulations of various interpretations of the movement of organisms in a bird flock, fish school, or ant colonies. In this paper, a new data visualization method, which was inspired by real birds behaviors, is proposed. In this method, each data pattern in the data set to be clustered is regarded as a piece of food and these data patterns will be sequentially tossed to a flock of birds on the ground. The flock of birds adjusts its physical movements to seek food. Individual members of the flock can profit from discoveries of all of other members of the flock during the search for food because an individual is influenced by the success of the best neighbor and tries to imitate the behavior of the best neighbor. Gradually, the flock of birds will be divided into several groups according to the distributions of the food. The formed groups will naturally correspond to the underlying data structures in the data set. However many practical data sets are consisted of high-dimensional data points; therefore, how to generalize the aforementioned idea to cluster high-dimensional data sets is a very demanding challenge. Since the Self-Organizing Map (SOM) algorithm can project high-dimensional data points into a low-dimensional space through a self-organizing procedure we decide to integrate the SOM algorithm with the foregoing swarm intelligence to propose a new data visualization algorithm d. We then name the new data visualization algorithm as the Swarm Intelligence-based SOM (SISOM) algorithm. The algorithm allows us to use our visualization to decide the numbers of clusters and then cluster the data set based on the estimated cluster number. Nine data sets are used to demonstrate the effectiveness of the proposed algorithm.
39
Wang, Kuo-min, and 王國銘. "Extending Structure Adaptive Self-Organizing Map for Classifying Mixed Data." Thesis, 2006. http://ndltd.ncl.edu.tw/handle/79691783262271689269.
Full textAbstract:
碩士國立雲林科技大學
資訊管理系碩士班
94
The self-organizing map (SOM) is a visualized technique, which has been extensively applied in data mining. The SOM can project high-dimensional data into low-dimensional space while preserving the original topological relation. However, traditional SOM fixes the structure of the map and can not dynamically expand neurons. Moreover, when used for classification traditional SOM usually obtained low accuracy. The structure adaptive self-organizing map (SASOM) solved the problem by dynamically splitting neurons to make all the data in a neuron have the same class label. However, SOM and SASOM can not reasonably deal with categorical values. In this paper, we propose a Generalized SASOM (GSASOM) to classify mixed, numeric and categorical, data, and to visualize the distribution of data with class labels. Our model integrates the ideas from SOM, GSOM, and SASOM, and overcomes their shortcomings. The experimental results confirm that our method maintains classification accuracy compared to that of SASOM. When categorical data are involved in the training data, GSASOM can faithfully preserve the topological order, and also can provide more useful visualized information.
40
Chung, Pei Ling, and 鍾佩陵. "Applying Self-Organizing Map in Clustering Protein Phosphorylation Sequence Data." Thesis, 2010. http://ndltd.ncl.edu.tw/handle/17423003007788194328.
Full textAbstract:
碩士長庚大學
資訊管理學系
98
The analysis of the surrounding sequences of protein phosphorylation sites is a very important topic in the protein and biology related research. However, the current data mining techniques applied on protein phosphorylation sites research mostly focus on building classification models to forecast the phosphorylation sites. This research applies cluster analysis in protein phosphorylation site sequences to assist biomedical researchers to filter through possible targets with reduced time and effort. Self-organizing map (SOM) is an often used cluster analysis method in biomedical data processing and with proven effectiveness. This thesis used the physical-chemical properties of amino acids and binary coding to encode protein phosphorylation site sequences; different topology of SOM is also used to explore the protein phosphorylation site surrounding sequences data which regulated by specific protein kinase. This research conducts cluster analyses on the PKA group kinase-related and CK2 group kinase-related protein phosphorylation site sequence data. The cluster analysis results of two data coding methods were evaluated, compared, and analyzed by five evaluation indicators and two color representation methods, and the results show that the cluster analysis on data coding using the physical-chemical properties of amino acid can better separate amino acid sequence of similar properties.
41
Hung, Hui-Fen, and 洪慧芬. "TAIEX Prediction Based on Self-Organizing Map and Grey System." Thesis, 2009. http://ndltd.ncl.edu.tw/handle/14081579422711242295.
Full textAbstract:
博士國立臺灣科技大學
電機工程系
97
Time series prediction is a difficult task. The prediction of stock price trends is particularly difficult because the volatility of the stock market index is affected by human, economic, political, technical and industrial factors. Among the currently used prediction mechanisms, neural networks show the best performance. Neural networks are nonlinear adaptive models and have the ability to approximate complex relations which may not have a prespecified form. A self-organizing map (SOM) is an unsupervised neural network that uses the similarity of high-dimensional data in a two-dimensional or one-dimensional coordinate space to facilitate data classification. GM(1,1) is a basic model for grey prediction and is capable of providing system predictions under the constraints of uncertainty and imperfect information. This paper presents a series of integrated time series prediction mechanisms. In addition to Grey SOM based on GRG (GSOMGRG) and Grey Fourier SOM based on GRG (GFSOMGRG), we explain the concepts of Two-stage GSOMGRG and Two-stage GFSOMGRG, which respectively predict and estimate the performance of models of time series data. The traditional SOM and Grey Relational Grade (GRG) of grey theory are combined into a new clustering mechanism, SOMGRG. A period of time series data is regarded as a pattern, and patterns are clustered with SOMGRG, together with the pre- and post-processing mechanisms of time series data, so that the accuracy of the predicted values is further enhanced. This prediction mechanism first uses GM(1,1) to predict the time series, and then the differencing data is regarded as forming patterns. The patterns are then clustered with GRG into a SOM structure. Then, the predicted values of the same type of data are modified to obtain better values. In the GM(1,1) predicting process, GFSOMGRG uses Fourier residual correction to increase the GM (1,1) prediction accuracy. For developing SOMGRG clustering patterns, a two-stage mechanism performs clustering twice in order to improve the accuracy of SOMGRG clustering. The experimental results show that Two-stage GFSOMGRG has better forecasting ability compared to other models and can help investors predict future stock price indices and grasp profit-making opportunities.
42
Chang, Chia-Chuang, and 張家銓. "Improved Self-organizing Linear Output Map for Reservoir Inflow Forecasting." Thesis, 2009. http://ndltd.ncl.edu.tw/handle/25754392471904807349.
Full textAbstract:
碩士國立臺灣大學
土木工程學研究所
97
Based on self-organizing linear output map (SOLO), effective hourly reservoir inflow forecasting models are proposed. As compared with back-propagation neural network (BPNN) which is the most frequently used conventional neural network (NN), SOLO has four advantages: (1) SOLO has better generalization ability; (2) the architecture of the SOLO is simpler; (3) SOLO is trained much more rapidly, and (4) SOLO could provide features that facilitate insight into underlying processes. An application is conducted to clearly demonstrate these four advantages. The results indicate that the SOLO model is more well-performed and efficient than the existing BPN-based models. To further improve the peak inflow forecasting, SOLO with data preprocessing named ISOLO is also proposed. The comparison between SOLO and ISOLO confirms the significant improvement in peak inflow forecasting. The proposed model is recommended as an alternative to the existing models. The proposed modeling technique is also expected to be useful to support reservoir operation systems.
43
Chen, Heng-Yu, and 陳恆裕. "An Efficient Self-Organizing Map Algorithm Based on Reference Point." Thesis, 2006. http://ndltd.ncl.edu.tw/handle/04399477595600384164.
Full textAbstract:
碩士國立中興大學
資訊科學系所
94
The self-organizing map (SOM) is an excellent mechanism for data mining. It has been used as a tool for mapping high-dimensional data into a two- (or three-) dimensional feature map. Despite its successes in practical applications, SOM suffers from some drawbacks such as trial-and-error method for searching a neighborhood preserving feature map. In this paper, we present an efficient self-organizing map algorithm to improve the performance of SOM. We use an efficient self-organizing map algorithm based on reference point and two threshold values .We use a threshold value as the search boundary which is used to search for the Best-Matching Unit (BMU) via input vector. Another threshold value is used as the search boundary in which the BMU finds its neighbors. Moreover, we propose a new method to lower the number of computations required when the Efficient Initialization Scheme for the Self-organizing Feature Map Algorithm is applied. The method reduce the time complexity form O(n2) to O(n) in the steps of finding the initial neurons. We ran our algorithm based on the data set from Yeast database and UCI KDD Archive to illustrate the performance improvement of the proposed method. In the experiment, the execution time of the original SOM algorithm is cut in half in our scheme. At the same time, the sum of squared-error distance in our scheme is also smaller than that of SOM. After achieving improvement of time complexity, this method is good enough to apply in the first-layer algorithm of the TWO LEVEL Based SOM.
44
Huang, Kuan-Wen, and 黃冠文. "Self-Organizing Map and Nonlinear Autoregression Networks for RegionalGroundwater Forecasting." Thesis, 2015. http://ndltd.ncl.edu.tw/handle/32628949080973269124.
Full textAbstract:
碩士淡江大學
水資源及環境工程學系碩士班
103
World climate becoming more extreme, department of water scarcity problem currently facing the world''s, Taiwan is limited by time and space uneven terrain and rainfall, each person assigned to the low rainfall is the world standard. How to preserve and recharge groundwater effectively has become an important issue.Groundwater has become an important water resource because of its low cost and easy extraction, often in the absence of sufficient surface water supply, it has become an important alternative water sources. The alluvial of the Zhuoshui River are good natural recharge areas of groundwater. Change of control and forecasting of groundwater, assist decision-making joint use and allocation management of surface water and groundwater reference. In this study, the study area is in the upstream mountain, upstream proximal-fan, midstream proximal-fan and downstream proximal-fan of Zhuoshui River. Collect the daily long-term (2000-2013) regional data sets and pre-processthe data of surface water and groundwater. Discussion groundwater aquifers of different districts and distribution and change, in non-hierarchical non-district "region mode" and "hierarchical partitioning mode" stratified-district of study area. The process is divided into build mode: data processing, SOM classification analysis, NARX. The results show that groundwater in the study area of classification, in 5X5 network is the most appropriate size. Available representative groundwater table, the amount of the spatial distribution of topography, and effective analysis of each neuron characteristics in agricultural water (irrigation and aquaculture water) at different times. NARX average groundwater level forecast model for the region quite excellent performance, R^2 are over 0.99 above. SOM-NARX mode groundwater variation prediction mode hierarchical partitioning of the region, in a hierarchical partitioning scheme outperformed the region''s performance mode, north of layering and zoning pattern by pattern outperformed the south coast of Zhuoshui River.
45
Lin, Shu-Han, and 林書漢. "Apply Extended Self-Organizing Map to Analyze Mixed-Type Data." Thesis, 2010. http://ndltd.ncl.edu.tw/handle/98921067198051589700.
Full textAbstract:
碩士雲林科技大學
資訊管理系碩士班
98
Mixed numeric and categorical data are commonly seen in nowadays corporate databases in which precious patterns may be hidden. Analyzing mixed-type data to extract the hidden patterns valuable to decision-making is therefore beneficial and critical for corporations to remain competitive. In addition, visualization facilitates exploration in the early stage of data analysis. In the paper, we present a visualized approach for analyzing multivariate mixed-type data. The proposed framework based on an extended self-organizing map allows visualized data cluster analysis as well as classification. We demonstrate the feasibility of the approach by analyzing two real-world datasets and compare our approach with other existing models to show its advantages.
46
Shih, Chi-Wei, and 石琢暐. "Information Hiding in Static Image Based on Self-Organizing Map." Thesis, 2008. http://ndltd.ncl.edu.tw/handle/c73zkx.
Full textAbstract:
碩士國立臺東大學
教育學系(所)
96
Originally, for studies in information hiding using vector quantization(VQ) techniques, data or images are encrypted before they are embedded into a cover image. However, using such a scheme is difficult to determine the capacity that a cover image can manage. In this study, an information hiding scheme using Self-Organizing Map and Least Significant Bit techniques is proposed to facilitate hiding unencrypted information in cover images. With the proposed scheme, the information hiding capacity of a cover image is deterministic; a cover image can embed information that its size is slightly less than the size of the cover image. Comparing with Linde-Buzo-Gray(LBG) algorithm applied to traditional VQ, the proposed scheme produces better quality of stego and secret images based on PSNR measurement. There are no significant quality differences between the proposed scheme and VQ, if secret information is in audio format.
47
Jheng, Jian-Jhong, and 鄭建忠. "Batch Training Algorithm for Mixed-Type Self-Organizing Map with a Fixed-Sized Map." Thesis, 2016. http://ndltd.ncl.edu.tw/handle/8839j2.
Full textAbstract:
碩士國立雲林科技大學
資訊管理系
104
The Kohonen’s Self-Organizing Map (SOM) is a well-known unsupervised learning algorithm in the visualization field. For the Self-Organizing Map (SOM), how to find out useful information or knowledge in real-world data is very important. In recent years, the SOM has had many variants that are improved and extended to the original model. For instance, the Generalized Self-Organizing Map (GenSOM) is able to process categorical data and mixed-type data as well. The Batch Generalized Self-Organizing Map (BatchGSOM) is batch and dynamic growth version of the GenSOM algorithm, which runs faster than the previous algorithms. However, if the real-world data is high-dimensional and has a large amount of data, many of neurons will be dynamically generated during the training process of BatchGSOM. As a result, the training time will increase due to the large amount of calculation. Therefore, in this study we propose a fixed-sized map version of batch, extended SOM so as to improve the performance. Experimental results indicate that the proposed approach outperforms the previous approaches in terms of training efficiency.
48
Wang, Wu-Han, and 王梧翰. "Exploring the Effective Evaluation Indices of Self-Organizing Map for Clustering Regional Flood Inundation Map." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/sdu4s7.
Full textAbstract:
碩士淡江大學
水資源及環境工程學系碩士班
106
Today, Artificial Intelligence is one of popular issues with many research topics and practical applications. The relative AI issues on the study of water resource management or flood forecast have become one of important topics. The purpose of this study is to propose the methodology to automatically build the Self-organizing maps (SOM) on clustering the flood spatial distribution. There are three major problems on building the SOM model; first one is the topological error, that is, any two neurons flip each other weights that makes the order of the topological map; second one is to the selection of the number of epochs. The training algorithm of SOM has two phases, ordering phase and convergent phase. Hence, these two phases have the different number of epochs and the number of epochs can influence the convergence; third one is to decide the optimal size. This study proposes two training strategies of the SOM models and takes Luermen Creek and Yenshui Creek located in Tainan, and Kemaman River located in Terengganu of Malaysia to investigate the convergence of the SOM models. The first strategy, called plan1, is to train the network in the ordering phase until the weights of the neurons have no obvious change, then transfer to the convergent phase and continue training the neurons until the weights have no obvious change. The second strategy, called plan2, is to rain the network in the ordering phase until the coverage rate of weights reaches 50%, then transfer to the convergent phase and continue training the same as the convergent phase of plan1. We use the flood simulation data of these three areas as the training data to build their own models. Through the different training strategy of plan1 and plan2, we can explore the influences of the ordering and convergent phases on building the SOM models. Through coverage rate, flip detector and five indices to compare the clustering results of the SOM clustering results. The coverage rate is defined as the difference of the cumulative distribution rates between maximum and minimum weights (neurons). The flip detector can check whether any two or more neurons flip each other weights or not and determine topological order correct or not. The clustering results of these three cases show that the number of epochs can influence the coverage rate and effectively improve the clustering quality. The larger number of epochs can get the larger coverage rate. The results show that plan2 can get convergent clustering results while plan1 occurs flip in Luermen Creek and Kemaman River. Hence plan2 is more suitable than plan1 for applying the SOM model on clustering the flood spatial distribution. Moreover, for comparison of the different size of the SOM models, the results demonstrate that the coverage rates of 3×3 model are smaller than those of 4×4 and 5×5 models, about 5%-10% less. That means 3×3 model cannot describe the characteristics of data as well as 4×4 and 5×5 models. The coverage rates of 4×4 and 5×5 models are almost the same, so the small models should be enough neurons to describe the data, that is, 4×4 is an appropriate size than other models. Hence, for choosing the size of topology map, the coverage rate is the great index to decide the optimal size.
49
Chun-Ming, Wang. "Integration of cluster analysis and discrimination analysis using self-organizing map." 2006. http://www.cetd.com.tw/ec/thesisdetail.aspx?etdun=U0001-2112200521430300.
Full text
50
Tsai, Wen-Cheng, and 蔡文誠. "Applying visualized self-organizing map to analyze digital camera product data." Thesis, 2008. http://ndltd.ncl.edu.tw/handle/08881492314112066779.
Full textAbstract:
碩士雲林科技大學
資訊管理系碩士班
96
The advent of modern information technology results in prosperous development of electronic commerce in the Internet so that Internet shopping is becoming more and more popular. However, many shoppers still face a lot of inconveniences nowadays when they conduct on-line shopping. The problems include the following：(1) There are too many products in an e-commerce shop store. (2) Similar products which have about the same functionally may have very different prices due to different brands.In the situation, it is essential for consumers to spend time comparing and choosing the product in order to acquire a product which matches his need with a reasonable price. However, manually comparing product information revealed on a website is tedious and costly. We plan to develop a framework of analyzing on-line products to support consumers effectively choosing appropriate commodity from the Internet. With the product of digital camera as an example, we collect product data including function of digital camera, product information, price and so on. They will be clustered according to clustering analysis by ViSOM from different aspects such as consumers, price, competitive strategy of manufacturers. The result of the experiment can help to choose a digital camera which fits the consumer’s need. In addition, we can provide effective strategy to manufactures of digital camera from different analysis aspects. In our experiments, we use the real digital camera datasets collected in the fourth season of 2007 from the Web to analyze. The preliminarily experimental results demonstrate that the proposed framework of data mining is feasible for online shopping in terms of seeking out a good buy for consumers and comparing product and pricing strategies for manufacturers. In the future, we hope to extend the analytic model further to find out more useful knowledge from the collected product dataset.