Home About us Contact
|
Home About us Contact | ||
|
|
||
Self-organizing Map (self-organizing + map)
Kinds of Self-organizing Map Selected AbstractsA study on two-stage self-organizing map and its application to clustering problemsELECTRICAL ENGINEERING IN JAPAN, Issue 1 2007Satoru Kato Abstract This paper presents a two-stage self-organizing map algorithm that we call two-stage SOM which combines Kohonen's basic SOM (BSOM) and Aoki's SOM with threshold operation (THSOM). In the first stage of two-stage SOM, we use BSOM algorithm in order to acquire topological structure of input data, and then we apply THSOM algorithm so that inactivated code vectors move to appropriate region reflecting the distribution of the input data. Furthermore, we show that two-stage SOM can be applied to clustering problems. Some experimental results reveal that two-stage SOM is effective for clustering problems in comparison with conventional methods. © 2007 Wiley Periodicals, Inc. Electr Eng Jpn, 159(1): 46,53, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.20268 [source] Strategies for preventing defection based on the mean time to defection and their implementations on a self-organizing mapEXPERT SYSTEMS, Issue 5 2005Young Ae Kim Abstract: Customer retention is a critical issue for the survival of any business in today's competitive marketplace. In this paper, we propose a dynamic procedure utilizing self-organizing maps and a Markov process for detecting and preventing customer defection that uses data of past and current customer behavior. The basic concept originates from empirical observations that identified that a customer has a tendency to change behavior (i.e. trim-out usage volumes) before eventual withdrawal and defection. Our explanatory model predicts when potential defectors are likely to withdraw. Two strategies are suggested to respond to the question of where to lead potential defectors for the next stage, based on anticipating when the potential defector will leave. Our model predicts potential defectors with little deterioration of prediction accuracy compared with that of the multilayer perceptron neural network and decision trees. Moreover, it performs reasonably well in a controlled experiment using an online game. [source] The Kohonen self-organizing map: an application to the study of strategic groups in the UK hotel industryEXPERT SYSTEMS, Issue 1 2001Bruce Curry This paper examines a neural network method known as the self-organizing map (SOM). The motivation behind the SOM is to transform the data to a two-dimensional grid of nodes while preserving its 'topological' structure. In neural network terminology this involves unsupervised learning. The nearest related statistical technique is cluster analysis. We employ the SOM in the task of identifying strategic groups of companies, using data which relate to the generic strategies suggested by Porter. Following identification of different groups of hotels with certain strategic emphases, the study investigates correlations between strategies followed and hotel performance. We compare and contrast the 'feature map' generated by the SOM with the results of a standard cluster analysis using the k-means method. The data also cover performance indicators and the results indicate that performance varies between strategic groups. [source] Time series forecasting by combining the radial basis function network and the self-organizing mapHYDROLOGICAL PROCESSES, Issue 10 2005Gwo-Fong Lin Abstract Based on a combination of a radial basis function network (RBFN) and a self-organizing map (SOM), a time-series forecasting model is proposed. Traditionally, the positioning of the radial basis centres is a crucial problem for the RBFN. In the proposed model, an SOM is used to construct the two-dimensional feature map from which the number of clusters (i.e. the number of hidden units in the RBFN) can be figured out directly by eye, and then the radial basis centres can be determined easily. The proposed model is examined using simulated time series data. The results demonstrate that the proposed RBFN is more competent in modelling and forecasting time series than an autoregressive integrated moving average (ARIMA) model. Finally, the proposed model is applied to actual groundwater head data. It is found that the proposed model can forecast more precisely than the ARIMA model. For time series forecasting, the proposed model is recommended as an alternative to the existing method, because it has a simple structure and can produce reasonable forecasts. Copyright © 2005 John Wiley & Sons, Ltd. [source] Electricity peak load forecasting with self-organizing map and support vector regressionIEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING, Issue 3 2006Shu Fan Non-member Abstract This paper aims to study the short-term peak load forecasting (PLF) by using Kohonen self-organizing maps (SOM) and support vector regression (SVR). We first adopt a SOM network to cluster the input data set into several subsets in an unsupervised learning strategy. Then, several SVRs for the next day's peak load are used to fit the training data of each subset in the second stage. In the numerical experiments, data of electricity demand from the New York Independent System Operator (ISO) are used to verify the effectiveness of the prediction for the proposed method. The simulation results show that the proposed model can predict the next day's peak load with a considerably high accuracy compared with the ISO forecasts. © 2006 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [source] Satellite image segmentation using hybrid variable genetic algorithmINTERNATIONAL JOURNAL OF IMAGING SYSTEMS AND TECHNOLOGY, Issue 3 2009Mohamad M. Awad Abstract Image segmentation is an important task in image processing and analysis. Many segmentation methods have been used to segment satellite images. The success of each method depends on the characteristics of the acquired image such as resolution limitations and on the percentage of imperfections in the process of image acquisition due to noise. Many of these methods require a priori knowledge which is difficult to obtain. Some of them are parametric statistical methods that use many parameters which are dependent on image property. In this article, a new unsupervised nonparametric method is developed to segment satellite images into homogeneous regions without any a priori knowledge. The new method is called hybrid variable genetic algorithm (HVGA). The variability is found in the variable number of cluster centers and in the changeable mutation rate. In addition, this new method uses different heuristic processes to increase the efficiency of genetic algorithm in avoiding local optimal solutions. Experiments performed on two different satellite images (Landsat and Spot) proved the high accuracy and efficiency of HVGA compared with another two unsupervised and nonparametric segmentation methods genetic algorithm (GA) and self-organizing map (SOM). The verification of the results included stability and accuracy measurements using an evaluation method implemented from the functional model (FM) and field surveys. © 2009 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 19, 199,207, 2009 [source] An artificial neural network based approach for online string matching/filtering of large databases,INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS, Issue 4 2010Tatiana Tambouratzis A novel online approach to exact string matching and filtering of large databases is presented. String matching/filtering is based on artificial neural networks and operates in two stages: initially, a self-organizing map retrieves the cluster of database strings that are most similar to the query string; subsequently, a harmony theory network compares the retrieved strings with the query string and determines whether an exact match exists. The similarity measure is configured to the specific characteristics of the database so as to expose overall string similarity rather than character coincidence at homologous string locations. The experimental results demonstrate foolproof, fast, and practically database-size independent operation that is especially robust to database modifications. The proposed approach is put forward for general-purpose (directory, catalogue, glossary search) as well as Internet-oriented (e-mail blocking, URL, username classification) applications. © 2010 Wiley Periodicals, Inc. [source] SOM-based estimation of climatic profilesINTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS, Issue 5 2006Tatiana Tambouratzis This article introduces a self-organizing map-based approach for estimating the climatic profile of locations of interest situated within an area of known morphology. The potential of the proposed methodology is illustrated on a number of locations within the Greek territory, and its superiority over other,customarily used as well as novel,climatic profile estimation methodologies is demonstrated and numerically evaluated. It is envisioned that, after further development, the proposed methodology can be employed for creating accurate climatic maps of areas of known morphology. © 2006 Wiley Periodicals, Inc. Int J Int Syst 21: 503,522, 2006. [source] Analyzing the 24-hour blood pressure and heart-rate variability with self-organizing feature mapsINTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS, Issue 1 2002G. Tambouratzis In this article, the self-organizing map (SOM) is employed to analyze data describing the 24-hour blood pressure and heart-rate variability of human subjects. The number of observations varies widely over different subjects, and therefore a direct statistical analysis of the data is not feasible without extensive pre-processing and interpolation for normalization purposes. The SOM network operates directly on the data set, without any pre-processing, determines several important data set characteristics, and allows their visualization on a two-dimensional plot. The SOM results are very similar to those obtained using classic statistical methods, indicating the effectiveness of the SOM method in accurately extracting the main characteristics from the data set and displaying them in a readily understandable manner. In this article, the relation is studied between the representation of each subject on the SOM, and his blood pressure and pulse-rate measurements. Finally, some indications are included regarding how the SOM can be used by the medical community to assist in diagnosis tasks. © 2002 John Wiley & Sons, Inc. [source] The analysis of motor vehicle crash clusters using the vector quantization techniqueJOURNAL OF ADVANCED TRANSPORTATION, Issue 3 2010Lorenzo Mussone Abstract In this paper, a powerful tool for analyzing motor vehicle data based on the vector quantization (VQ) technique is demonstrated. The technique uses an approximation of a probability density function for a stochastic vector without assuming an "a priori" distribution. A self-organizing map (SOM) is used to transform accident data from an N-dimensional space into a two-dimensional plane. The SOM retains all the original data yet provides an effective visual tool for describing patterns such as the frequency at which a particular category of events occurs. This enables new relationships to be identified. Accident data from three cities in Italy (Turin, Milan, and Legnano) are used to illustrate the usefulness of the technique. Crashes are aggregated and clustered crashes by type, severity, and along other dimensions. The paper includes discussion as to how this method can be utilized to further improve safety analysis. Copyright © 2010 John Wiley & Sons, Ltd. [source] Patternizing of impoundment impact (1985,2002) on fish assemblages in a lowland river using the Kohonen algorithmJOURNAL OF APPLIED ICHTHYOLOGY, Issue 3 2005T. Penczak Summary Impoundment impact on fish assemblage structure was investigated in the dammed middle course of the Warta River. A backwater site (AB) was located 2 km upstream of the Jeziorsko Reservoir, and a tailwater site (CD) 1.5 km downstream of the dam. Both sites were studied for 3 years in the pre-impoundment period (1985,1987) and 15 years after damming (1988,2002). Quantitative electrofishing in spring and autumn assured obtaining yearly average biomass for each population. Most of the data analysis aimed to assess the dam impact on the fish assemblage structure but other accompanying impacts such as discharge manipulations, revetment, different forms of engineering, and water quality improvement in the tailwater and backwater reaches were also discussed. The Kohonen algorithm (self-organizing map, SOM) was used for the analysis, and perfectly separated AB and CD samples into two clusters. Samples from the backwater (AB) proved that this reach of the Warta River had maintained its almost natural character and that fish assemblages had changed moderately, now occupying only five neighbouring hexagons out of a total of 16. In the tailwater (CD), however, because of considerable fluctuations in fish assemblages the SOM produced three subclusters, which engaged nine hexagons: (i) the pre-impoundment period (1985,1987, two hexagons); (ii) 7 years after the definite closure of dam sluices (1988,1994, five hexagons); and (iii) the past 8 years of sampling (1995,2002, two hexagons), when stabilization in the assemblage was observed. The SOM also definitely proved profound changes in fish assemblage composition: most lithophilous species declined and many phytolithophilous and phytophilous species became dominants, particularly in the tailwater site where downstream migration of 0+ of successfully spawned species from the reservoir took place. [source] Modeling and predicting binding affinity of phencyclidine-like compounds using machine learning methodsJOURNAL OF CHEMOMETRICS, Issue 1 2010Ozlem Erdas Abstract Machine learning methods have always been promising in the science and engineering fields, and the use of these methods in chemistry and drug design has advanced especially since the 1990s. In this study, molecular electrostatic potential (MEP) surfaces of phencyclidine-like (PCP-like) compounds are modeled and visualized in order to extract features that are useful in predicting binding affinities. In modeling, the Cartesian coordinates of MEP surface points are mapped onto a spherical self-organizing map (SSOM). The resulting maps are visualized using electrostatic potential (ESP) values. These values also provide features for a prediction system. Support vector machines and partial least-squares method are used for predicting binding affinities of compounds. Copyright © 2009 John Wiley & Sons, Ltd. [source] Organizing learning materials through hierarchical topic maps: an illustration through Chinese herb medicationJOURNAL OF COMPUTER ASSISTED LEARNING, Issue 6 2007B.-J. Shih Abstract This research aims to use hierarchical topic maps to compile digital learning material and to discuss its design and application possibilities. The system renders tremendous original assets and then embeds a self-organizing map (SOM) in the material database to produce topical learning materials, as in this case, an illustration through Chinese herb medication. It helps to demonstrate robust professional information as well as knowledge structures, and provides a customized and interactive learning dynamic to support both progressive and constructive learning styles. The paper first gives a detailed procedural description of the material construction, explains how topic map techniques were applied, and observes the implications and potentials of the technology to education. Both the technical and educational evaluations of using SOM topic maps in compilation of learning materials have resulted in positive feedback. SOM allows users to review the complete databank in structural hierarchical order, which provides comprehensive understanding of the entire set of learning materials, and also brings opportunities to users to discover knowledge related to their study area. [source] Fault detection and isolation in robotic manipulators via neural networks: A comparison among three architectures for residual analysisJOURNAL OF FIELD ROBOTICS (FORMERLY JOURNAL OF ROBOTIC SYSTEMS), Issue 7 2001Marco Henrique Terra In this article we discuss artificial neural networks-based fault detection and isolation (FDI) applications for robotic manipulators. The artificial neural networks (ANNs) are used for both residual generation and residual analysis. A multilayer perceptron (MLP) is employed to reproduce the dynamics of the robotic manipulator. Its outputs are compared with actual position and velocity measurements, generating the so-called residual vector. The residuals, when properly analyzed, provides an indication of the status of the robot (normal or faulty operation). Three ANNs architectures are employed in the residual analysis. The first is a radial basis function network (RBFN) which uses the residuals of position and velocity to perform fault identification. The second is again an RBFN, except that it uses only the velocity residuals. The third is an MLP which also performs fault identification utilizing only the velocity residuals. The MLP is trained with the classical back-propagation algorithm and the RBFN is trained with a Kohonen self-organizing map (KSOM). We validate the concepts discussed in a thorough simulation study of a Puma 560 and with experimental results with a 3-joint planar manipulator. © 2001 John Wiley & Sons, Inc. [source] Image Segmentation and Bruise Identification on Potatoes Using a Kohonen's Self-Organizing MapJOURNAL OF FOOD SCIENCE, Issue 7 2005Thierry Marique ABSTRACT: Potato quality includes a low incidence of colored bruises resulting from bad storage or manipulation practices. We developed a procedure to process and segment potato images by using Kohonen's self-organizing map. Anomalous regions could be distinguished on 3 potato varieties. Bruises that were very dissimilar in appearance were correctly identified, and some particular defects such as green spots could be located as well. [source] WELL LOG CALIBRATION OF KOHONEN-CLASSIFIED SEISMIC ATTRIBUTES USING BAYESIAN LOGICJOURNAL OF PETROLEUM GEOLOGY, Issue 4 2001M. T. Taner We present a new method for calibrating a classified 3D seismic volume. The classification process employs a Kohonen self-organizing map, a type of unsupervised artificial neural network; the subsequent calibration is performed using one or more suites of well logs. Kohonen self-organizing maps and other unsupervised clustering methods generate classes of data based on the identification of various discriminating features. These methods seek an organization in a dataset and form relational organized clusters. However, these clusters may or may not have any physical analogues in the real world. In order to relate them to the real world, we must develop a calibration method that not only defines the relationship between the clusters and real physical properties, but also provides an estimate of the validity of these relationships. With the development of this relationship, the whole dataset can then be calibrated. The clustering step reduces the multi-dimensional data into logically smaller groups. Each original data point defined by multiple attributes is reduced to a one- or two-dimensional relational group. This establishes some logical clustering and reduces the complexity of the classification problem. Furthermore, calibration should be more successful since it will have to consider less variability in the data. In this paper, we present a simple calibration method that employs Bayesian logic to provide the relationship between cluster centres and the real world. The output will give the most probable calibration between each self-organized map node and wellbore-measured parameters such as lithology, porosity and fluid saturation. The second part of the output comprises the calibration probability. The method is described in detail, and a case study is briefly presented using data acquired in the Orange River Basin, South Africa. The method shows promise as an alternative to current techniques for integrating seismic and log data during reservoir characterization. [source] Long-term radio behaviour of GPS sources and candidatesASTRONOMISCHE NACHRICHTEN, Issue 2-3 2009M. Tornikoski Abstract This paper is a summary of the work that our group has done (and recently published in several papers) on long-term radio variability of GPS sources. We have studied the long-term (up to 30 years) variability of GPS sources and candidates, with emphasis on the high-frequency radio domain. Our data sets show that only a relatively small number of these sources retain their convex spectra when they are monitored densely and for long periods of time. The current GPS samples are especially contaminated by small, beamed blazar-type sources. Also the remaining population with consistently convex GPS-type spectra seems to be heterogeneous, falling into several subpopulations when their observed properties are used for clustering them through a self-organizing map (© 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim) [source] Winner-relaxing and winner-enhancing Kohonen maps: Maximal mutual information from enhancing the winnerCOMPLEXITY, Issue 4 2003Jens Christian Claussen Abstract The magnification behavior of a generalized family of self-organizing feature maps, the winner relaxing and winner enhancing Kohonen algorithms is analyzed by the magnification law in the one-dimensional case, which can be obtained analytically. The winner-enhancing case allows to achieve a magnification exponent of one and therefore provides optimal mapping in the sense of information theory. A numerical verification of the magnification law is included, and the ordering behavior is analyzed. Compared to the original self-organizing map and some other approaches, the generalized winner enforcing algorithm requires minimal extra computations per learning step and is conveniently easy to implement. © 2003 Wiley Periodicals, Inc. [source] Zero-sequence-based relaying technique for protecting power transformers and its performance assessment using unsupervised learning ANNEUROPEAN TRANSACTIONS ON ELECTRICAL POWER, Issue 2 2006Guzmán Díaz Abstract In this paper a simple and robust new relaying technique for protecting transformers from internal winding faults is proposed. Based on the measurement of zero sequence current inside a delta winding, the technique greatly simplifies the conventional differential relaying arrangement when a delta winding is available. Despite the number of windings of the transformer and the location of the fault, only measurement of induced zero sequence current within the delta winding is needed. Since the proposed technique has been shown to be prone to generate false pick-up signals during inrush, a simple restraining criterion is proposed and analysed. Additionally, use of projection techniques based on self-organizing maps (SOM) is proposed in this paper as a valuable tool for analysing multivariable data which are generated from the huge number of possible combinations existing between switching instant and fault location. Finite element simulations and laboratory tests have been combined into SOM to validate the proposed relaying technique and the restraining criterion. Copyright © 2005 John Wiley & Sons, Ltd. [source] Strategies for preventing defection based on the mean time to defection and their implementations on a self-organizing mapEXPERT SYSTEMS, Issue 5 2005Young Ae Kim Abstract: Customer retention is a critical issue for the survival of any business in today's competitive marketplace. In this paper, we propose a dynamic procedure utilizing self-organizing maps and a Markov process for detecting and preventing customer defection that uses data of past and current customer behavior. The basic concept originates from empirical observations that identified that a customer has a tendency to change behavior (i.e. trim-out usage volumes) before eventual withdrawal and defection. Our explanatory model predicts when potential defectors are likely to withdraw. Two strategies are suggested to respond to the question of where to lead potential defectors for the next stage, based on anticipating when the potential defector will leave. Our model predicts potential defectors with little deterioration of prediction accuracy compared with that of the multilayer perceptron neural network and decision trees. Moreover, it performs reasonably well in a controlled experiment using an online game. [source] Electricity peak load forecasting with self-organizing map and support vector regressionIEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING, Issue 3 2006Shu Fan Non-member Abstract This paper aims to study the short-term peak load forecasting (PLF) by using Kohonen self-organizing maps (SOM) and support vector regression (SVR). We first adopt a SOM network to cluster the input data set into several subsets in an unsupervised learning strategy. Then, several SVRs for the next day's peak load are used to fit the training data of each subset in the second stage. In the numerical experiments, data of electricity demand from the New York Independent System Operator (ISO) are used to verify the effectiveness of the prediction for the proposed method. The simulation results show that the proposed model can predict the next day's peak load with a considerably high accuracy compared with the ISO forecasts. © 2006 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [source] Consensus between GCM climate change projections with empirical downscaling: precipitation downscaling over South AfricaINTERNATIONAL JOURNAL OF CLIMATOLOGY, Issue 10 2006B. C. Hewitson Abstract This paper discusses issues that surround the development of empirical downscaling techniques as context for presenting a new approach based on self-organizing maps (SOMs). The technique is applied to the downscaling of daily precipitation over South Africa. SOMs are used to characterize the state of the atmosphere on a localized domain surrounding each target location on the basis of NCEP 6-hourly reanalysis data from 1979 to 2002, and using surface and 700-hPa u and v wind vectors, specific and relative humidities, and surface temperature. Each unique atmospheric state is associated with an observed precipitation probability density function (PDF). Future climate states are derived from three global climate models (GCMs): HadAM3, ECHAM4.5, CSIRO Mk2. In each case, the GCM data are mapped to the NCEP SOMs for each target location and a precipitation value is drawn at random from the associated precipitation PDF. The downscaling approach combines the advantages of a direct transfer function and a stochastic weather generator, and provides an indication of the strength of the regional versus stochastic forcing, as well as a measure of stationarity in the atmosphere,precipitation relationship. The methodology is applied to South Africa. The downscaling reveals a similarity in the projected climate change between the models. Each GCM projects similar changes in atmospheric state and they converge on a downscaled solution that points to increased summer rainfall in the interior and the eastern part of the country, and a decrease in winter rainfall in the Western Cape. The actual GCM precipitation projections from the three models show large areas of intermodel disagreement, suggesting that the model differences may be due to their precipitation parameterization schemes, rather than to basic disagreements in their projections of the changing atmospheric state over South Africa. Copyright © 2006 Royal Meteorological Society. [source] Organizational emergence in networked collaborationINTERNATIONAL JOURNAL OF COMMUNICATION SYSTEMS, Issue 7 2002Ari-Pekka Hameri Abstract Research on complex adaptive systems has generated several conceptual parables to explain systems with emergent behaviour. One prominent use for terms such as self-organization, evolutionary trajectories, co-evolution and punctuated equilibrium has been in understanding human organizations. In such systems, emergent behaviour is demonstrated in novel structures, processes and spin-offs that cannot be explained just by studying single components of the organization and the intelligence embedded in them. Instead of solely exploiting the qualitative explanatory power of the evolutionary concepts, this paper focuses also on quantitative methods to track emergent behaviour in a globally distributed, constantly fluctuating and highly networked project organization. The underlying case is that of CERN (CERN, the European Laboratory for Particle Physics, has its headquarters in Geneva. At present, its Member States are Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Italy, Netherlands, Norway, Poland, Portugal, the Slovak Republic, Spain, Sweden, Switzerland and the United Kingdom. Israel, the Russian Federation, Turkey, Yugoslavia (status suspended after the UN embargo, June 1992), the European Commission and UNESCO have observer status.) and its decade long accelerator project, which strongly relies on electronic communication and networking to achieve its major objectives due to be accomplished by the year 2006. By using time series and self-organizing maps to analyse the global interaction among project groups and individuals the paper provides new insight to the understanding of emergent behaviour in human organizations. The key result of the study concerns the rigid deep structure of each case organization that seems to remain intact for the duration of the whole project. Copyright © 2002 John Wiley & Sons, Ltd. [source] Psychographic clustering of blood donors in Egypt using Kohonen's self-organizing mapsINTERNATIONAL JOURNAL OF NONPROFIT & VOLUNTARY SECTOR MARKETING, Issue 2 2010Mohamed M. Mostafa Blood donation has historically been marketed as one of the purest examples of altruistic or pro-social behavior. The recruitment of blood donors is, however, a challenge as transfusion centers worldwide strive to attract and retain blood donors. This study uses self-organizing maps (SOM) to examine the effect of various psychographic and cognitive factors on blood donation in Egypt. SOM is a machine learning method that can be used to explore patterns in large and complex datasets for linear and nonlinear patterns. The results show that major variables affecting blood donation are related to altruistic values, perceived risks of blood donation, blood donation knowledge, attitudes toward blood donation, and intention to donate blood. The study also shows that SOM models are capable of improving clustering quality while extracting valuable information from multidimensional data. Copyright © 2009 John Wiley & Sons, Ltd. [source] WELL LOG CALIBRATION OF KOHONEN-CLASSIFIED SEISMIC ATTRIBUTES USING BAYESIAN LOGICJOURNAL OF PETROLEUM GEOLOGY, Issue 4 2001M. T. Taner We present a new method for calibrating a classified 3D seismic volume. The classification process employs a Kohonen self-organizing map, a type of unsupervised artificial neural network; the subsequent calibration is performed using one or more suites of well logs. Kohonen self-organizing maps and other unsupervised clustering methods generate classes of data based on the identification of various discriminating features. These methods seek an organization in a dataset and form relational organized clusters. However, these clusters may or may not have any physical analogues in the real world. In order to relate them to the real world, we must develop a calibration method that not only defines the relationship between the clusters and real physical properties, but also provides an estimate of the validity of these relationships. With the development of this relationship, the whole dataset can then be calibrated. The clustering step reduces the multi-dimensional data into logically smaller groups. Each original data point defined by multiple attributes is reduced to a one- or two-dimensional relational group. This establishes some logical clustering and reduces the complexity of the classification problem. Furthermore, calibration should be more successful since it will have to consider less variability in the data. In this paper, we present a simple calibration method that employs Bayesian logic to provide the relationship between cluster centres and the real world. The output will give the most probable calibration between each self-organized map node and wellbore-measured parameters such as lithology, porosity and fluid saturation. The second part of the output comprises the calibration probability. The method is described in detail, and a case study is briefly presented using data acquired in the Orange River Basin, South Africa. The method shows promise as an alternative to current techniques for integrating seismic and log data during reservoir characterization. [source] Supporting user-subjective categorization with self-organizing maps and learning vector quantizationJOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY, Issue 4 2005Dina Goren-Bar Today, most document categorization in organizations is done manually. We save at work hundreds of files and e-mail messages in folders every day. While automatic document categorization has been widely studied, much challenging research still remains to support user-subjective categorization. This study evaluates and compares the application of self-organizing maps (SOMs) and learning vector quantization (LVQ) with automatic document classification, using a set of documents from an organization, in a specific domain, manually classified by a domain expert. After running the SOM and LVQ we requested the user to reclassify documents that were misclassified by the system. Results show that despite the subjective nature of human categorization, automatic document categorization methods correlate well with subjective, personal categorization, and the LVQ method outperforms the SOM. The reclassification process revealed an interesting pattern: About 40% of the documents were classified according to their original categorization, about 35% according to the system's categorization (the users changed the original categorization), and the remainder received a different (new) categorization. Based on these results we conclude that automatic support for subjective categorization is feasible; however, an exact match is probably impossible due to the users' changing categorization behavior. [source] | ||||||||||||||||||||||||||||||||||