Learn more about Viscovery products
More information
Viscovery in scientific articles
Learn how Viscovery products are used all over the world to solve typical data mining problems with self-organizing maps. Viscovery has been referenced in numerous publications, some of which are collected here for the following types of applications:
- Biology and life sciences
- Medical and clinical research
- Analysis of social and public systems
- Text mining and document classification
- Business intelligence
- Data mining trends and technologies
- Process optimization and engineering
- Market and finance analytics
- Web analytics
- Image and speech recognition
Biology and life sciences
Differences in the metabolite profiles of spinach (Spinacia oleracea L.) leaf in different concentrations of nitrate in the culture solution
Okazaki, Oka, Shinano, Osaki and Takebe (2008) Plant Cell Physiology 49:170-177
Metabolite profiling using gas chromatography-mass spectrometry (GC-MS) was used to evaluate the effect of N levels on spinach tissue, comparing two cultivars that differed in their ability to use N. SOM analysis using SOMine was used to describe changes in the metabolites of mature spinach leaves. Both PCA and SOM revealed that metabolites could be broadly divided into two types, correlating either positively or negatively with plant N content. The simple and co-coordinated metabolic stream, containing both general and spinach-specific aspects of plant N content, will be useful in future research on such topics as the detection of environmental effects on spinach through comprehensive metabolic profiling.
Sensor combination and chemometric modelling for improved process monitoring in recombinant E. coli fed-batch cultivations
Franz Clementschitsch, Kern Jürgen, Pötschacher Florentina and Bayer Karl, Journal of Biotechnology, Volume 120, Issue 2, 4 November 2005, Pages 183-196
The key objective for the optimization of recombinant protein production in bacteria is to maximize the exploitation of the host cell's synthesis potential. Since there are no reliable online-sensors for key variables, it is necessary to relate available online signals to process variables by mathematical models. To improve chemometric modeling of process variables, dielectric spectroscopy and a multi-wavelength online fluorescence sensor for two-dimensional fluorescence spectroscopy were applied in a series of recombinant Escherichia coli fed-batch cultivations applying two different process operation states. Chemometric modeling of key process variables with two different modeling techniques showed that this sensor combination greatly improved the estimation (i.e., reduce error magnitude) of process variables in recombinant E. coli cultivations, thereby enhancing process monitoring capabilities.
Individuality of wing patterning in Giant honey bees (Apis laboriosa)
Gerald Kastberger, Sarah Radloff and Gerhard Kranner (2003) Apidologie 34, 311-318
This study investigated whether individual worker bees of a single Apis laboriosa colony can be re-identified by their wing patterns alone. Re-identification was carried out by SOM reclassification and conventional discriminant analysis (DA) using the protocols of recognition (data for training and testing the model are equal or slightly modified by white noise), and prediction (test data are unknown to the model). SOM recognition of wing shaping was found to be more robust than that resulting from DA. The SOM prediction capacity was tested using four test-training data ratios and reached 90% under a two-step reclassification protocol.
Breeding rubis cultivars for high anthocyanin content and high antioxidant capacity
McGhie, Hall, Ainge and Mowat (2002) ISHS Acta Horticulturae 585: VIII International Rubus and Ribes Symposium
Anthocyanin content and antioxidant activity from HortResearch Rubus clones were assessed a diverse range of anthocyanins and total anthocyanin content was observed. These data could be used to improve commercial production of high-health Rubus crops with significantly higher anthocyanin content and antioxidant capacity than found in existing cultivars.
Visualization of multiple influences on ocellar flight control in giant honeybees with the data-mining tool Viscovery SOMine
Kastberger G; Kranner G (2000) Behav Res Methods Instrum Comput; 32(1):157-68
Viscovery SOMine was used to analyze and visualize multiple influences of the ocellar system on free-flight behavior in giant honeybees. Occlusion of ocelli will affect orienting reactivities in relation to flight target, level of disturbance, and position of the bee in the flight chamber; it will induce phototaxis and make orienting imprecise and dependent on motivational settings. Ocelli permit the adjustment of orienting strategies to environmental demands by enforcing abilities such as centering or flight kinetics and by providing independent control of posture and flight course.
Visualising spatial patterns in fruit quality and productivity of persimmon orchards using self organising maps
Mowat (2000) Presented at SIRC 2000 - The 12th Annual Colloquium of the Spatial Information Research Centre University of Otago, Dunedin, New Zealand, December 10-13th 2000
Fruit quality and productivity datasets, obtained over two seasons from 24 New Zealand persimmon orchards, were analyzed. SOMine was used to construct a 2000 node self organizing map (Kohonen, 1997) from input features (latitude, longitude, and growing region) obtained from each orchard replicate. By overlaying fruit quality and tree productivity attributes as component planes, spatial patterns between orchards could be observed. In addition, climatic data from regional meteorological stations was associated with the map.
Medical and clinical research
Anomaly Detection in Emergency Call Data – the First Step to the Intelligent Emergency Call System Management
Klement, Snášel (2009) International Conference on Intelligent Networking and Collaborative Systems (INCoS), Nov. 4-6th 2009, Barcelona, Spain
Past experiences in emergency calls are exploited to create a feedback to the emergency call taking process and a base for management decision support in order to improve the emergency response agencies organization and effectiveness.
Viscovery SOMine was used to visualize the emergency call taking information system database characteristics and discover further knowledge therein.
Combining data mining and case-based reasoning for intelligent decision support for pathology ordering by general practitioners
Zhuang, Churilov, Burstein, Sikaris (2009) European Journal of Operational Research, Volume 195, Issue 3, June 16th 2009, Pages 662-675
In this paper a novel methodology for integrating data mining and case-based reasoning for decision support for pathology ordering is proposed. It is demonstrated how this methodology can facilitate intelligent decision support that is both patient-oriented and deeply rooted in practical peer-group evidence. It is illustrated how knowledge extracted through data mining with Kohonen’s self-organizing maps constitutes the base that, with further assistance of the modern data visualization tool Viscovery SOMine and on-line processing interfaces, can facilitate more informed evidential decision making by doctors in the area of pathology ordering.
Using supervised and unsupervised techniques to determine groups of patients with different doctor-patient stability
Siew, Churilov, Smith-Miles and Sturmberg (2008) Lecture Notes in Computer Science
Similarities between any groupings found between unsupervised classification (SOFM) using SOMine and supervised (Classification and Regression Trees - CART) were compared and used to identify insights into factors associated with doctor-patient stability. Both methods resulted in many similar groupings indicating that self perceived health and age are important indicators of stability. Profiles of patients that are at risk were identified.
Combining data mining and discrete event simulation for a value-added view of a hospital emergency department
Ceglowski, Churilov and Wasserthiel (2007) J. Operational Res. Society ISSN: 0160-5682
Treatments given to emergency patients were clustered using SOMine. Analysis revealed treatments related to injury (e.g., tetanus injections, dressings, sutures) and treatments related to illness (e.g., arterial blood gases, echocardiograms, and intravenous drug infusion.
An Investigation of Emergency Department Overcrowding using Data Mining and Simulation: A Patient Treatment Type Perspective
Ceglowski, (2006) PhD Thesis, Monash University, Australia
In order to analyze the problem of overcrowding in emergency departments, homogenous clusters of patient treatment with similar activities were identified. Techniques from the dissociated methods of Data Mining and Management Science were combined within the hypothesis / experimentation framework of Scientific Method.
Viscovery SOMine has been used for discovery of patient treatment patterns. The clusters were combined with patient urgency and disposition to create “patient treatment types” that were tracked through the emergency department.
Knowledge Discovery through Mining Emergency Department Data
Ceglowski, Churilov and Wasserthiel (2005) Proceedings of the 38th Annual Hawaii International Conference on System Sciences 2005
The complexity of hospital emergency department operations limits comprehension and inhibits efforts to improve efficiency. Attempts have been made to reduce the complexity by streaming patients into similar classes of treatment or grouping them into similar cases. These have not successfully modeled the treatment of patients. Data mining techniques were employed to reduce the complexity of hospital emergency department operations by streaming patients into similar classes of treatment or grouping them into similar cases.
Viscovery SOMine was used to generate data models. The combination of a process philosophy with data mining resulted in the discovery of definitive “treatment pathways”. These pathways comprehensively model treatment of patients.
Analysis of hippocampal atrophy in alcoholic patients by a Kohonen feature map
Kurth et al (2004) Clinical Neuroscience and Neuropathology Neuroreport, 15(2):367-371
The correlation of hippocampal volume with homocysteine, folate, vitamin B12 and B6 in alcoholic patients and healthy controls was examined by applying a Kohonen feature map (KFM) and conventional statistics. KFM proved to be a sensitive tool for visualization of statistical correlations in data sets even if no further statistical information is available.
Using data mining techniques to identify groups of patients with different consultation satisfaction in general practice.
Siew, Churilov, Smith and Sturmberg (2004) ICOTA6 2004
SOMine was used to identify groups of patients with different levels of satisfaction. Results showed that doctor-patient communication is the most important variable in predicting satisfaction. Other important factors are patients’ knowledge of their doctor, doctor-patient stability, patient’s age, and consultation difficulty and length. This study has indicated profiles of patients that put them at risk of poor satisfaction which might be useful to the practicing doctors.
Emosys: a system to study hematological diseases
Starita, Rossi, F Caracciolo, and Petrini (2004) Proceeding (417) Biomedical Engineering 2004
The web-based Emosys system was developed both for managing and studying hematological diseases. The system is made of independent modules for the clinical management, for the collection of data, and for research on these diseases. Data mining techniques based on neural networks were used to confirm known results and/or to find interesting, unknown relationships among the data. The network was trained and the resulting map was graphically analyzed with the Viscovery SOM. The system is now under validation at the clinical site.
A neural clustering approach to iso-resource grouping for acute healthcare in Australia
Siew, Smith, Churilov and Ibrahim (2002) Proceedings of the 35th Hawaii International Conference on System Sciences 2002
The Case Mix funding formula is the most widely used approach for classifying patients according to diagnostic related groups (DRGs). Although it is clinically meaningful, experience suggests that DRG groupings do not necessarily present a sound basis for relevant knowledge generation. An alternative grouping of the patients based on a neural clustering approach is proposed, which generates homogeneous groups of patients with similar resource utilization characteristics. Features of the data and the dependencies between the variables were identified and evaluated from the SOMine map.
Clinical-pathological classification of glioblastomas investigated by a non-supervised neural network
Iglesias-Rozas, Camara and Schwemmle (2000) Electronic J. of Pathology and Histology 6:06 ISSN 0948-0382
Using a variant of unsupervised neural networks (Self-Organizing Maps, SOM), the ability to reproduce a clinical-pathological classification of patients with glioblastomas was examined. This SOM provides a powerful means to visualize and analyze complex data sets without prior statistical knowledge and allows a specific visual evaluation of new treatments and a more effective comparison with established tumor management.
Light microscope heterogeneity of human glioblastomas investigated with an unsupervised neural network (SOM)
Iglesias-Rozas and Grieshaber (2000) Electronic J Pathology and Histology 6:4/2000 p02
As an alternative to statistical evaluation of histological variability of glioblastomas, 1266 human glioblastomas were investigated to discover whether they can be correctly classified using SOMs generated with SOMine. Five clusters of glioblastomas with a maximum significance were found. A useful classification, comparable to the classification suggested by the World Health Organization, as well as the visualization of multidimensional histological features of human glioblastomas was achieved. The data can be used to improve patient management.
Analysis of social and public systems
Toward optimal calibration of the SLEUTH land use change model
Dietzel and Clarke (2007) Transactions in GIS 11:29-45
SLEUTH is a computational simulation model that uses adaptive cellular automata to simulate the way cities grow and change their surrounding land uses. SOMine was used to generate a SOM for reducing data, to pursue the isolation of the best parameter sets, and to indicate which of the existing 13 calibration metrics used in SLEUTH are necessary to arrive at the optimum. A new metric is proposed for increasing the value in future SLEUTH applications.
Operationalising multidimensional concepts of chronic poverty: an exploratory spatial analysis
Mehta, Panigrahi and Sivramkrishna (2006) Chronic Poverty and Development Policy in India, Aasha Kapur Mehta, Andrew Shepherd ISBN 0761934642
SOMine was used for the K-SOM analysis was used to detect spatial inequalities at all levels of disaggregation - between countries, states, regions, districts, blocks and even within cities, towns and villages. The nature and extent of these inequalities varies with choice of indicator and geographical space over which comparisons are made.
Towards fair ranking of Olympics achievements: the case of Sydney 2000
Churilov and Flitman (2006) Computers Operations Res. 33:2057-2082
An objective impartial system of analysis of the Olympic results, which the majority of participating countries would agree upon, is analyzed by discussing different ways of ranking the performance of participating countries at Sydney 2000 Olympic Games. The unsupervised data mining technique of self-organizing maps was used to group the participating countries into homogenous clusters. The Data Envelopment Analysis (DEA)-based model was then used for producing a new ranking of participating teams acceptable as “fair” by the majority of participants.
Self-organising map methods in integrated modelling of environmental and economic systems
Shanmuganathan, Sallis, Buckeridge in Environmental Modelling & Software, Volume 21, Issue 9, September 2006, Pages 1247-1256
The paper elaborates on how self-organizing map (SOM) methodologies (on the basis of Viscovery SOMine) within the connectionist paradigms of artificial neural networks (ANNs) could be applied to disparate data analysis at two different scales of environmental and economic systems: regional (using river water quality monitoring data to evaluate ecosystem response to human influence) and global (for modeling of environmental and economic system data and trade-off analysis) within an integrated framework to inform sustainable environment management.
Segmenting the market of West Australian senior tourists using an artificial neural network
Kim, Wie, Ruys in Tourism Management, Volume 24, Issue 1, February 2003, Pages 25-34
Using neural networks is one way to determine what trade-offs older travelers make as they decide their travel plans. This paper presents a descriptive analysis of neural network methodology and provides a research technique that assesses the weighting of different attributes and uses an unsupervised neural network model created by Viscovery SOMine to describe a consumer-product relationship. The model is nonlinear and does not require the same restrictive assumptions about the relationship between the independent variables and dependent variables..
An economic analysis of government transfers with Viscovery SOMine (Japanese language)
Aiko, Yu, Hiroshi, and Satomi (2000) Faji Shisutemu Shinpojiumu Koen Ronbunshu 16: 403-404
Using Viscovery SOMine, we analyzed financial transfers from the central government to local governments.
Self-Organizing Patterns in World Poverty using multiple indicators of poverty, repression and corruption
Deboeck, G. (2000), Neural Network World, Vol. 10, No. 1-2, p. 239-254
This paper maps world poverty based on multi-dimensions of poverty. These global maps are based on a well-established neural network algorithm implemented in the software tool Viscovery SOMine. They show world poverty based on similarity and dissimilarity in poverty structures.
Text mining and document classification
Text mining of medical records for radiodiagnostic decision-making
Claster, Shanmuganathan and Ghotbi (2008) J Computers 3:1
Radiology department records of children who had undergone a CT scan procedure at Nagasaki University Hospital in the year 2004 were analyzed using SOMine. Keywords were identified with a significance value within the narratives of the medical records that could predict and thereby lower the number of unnecessary CT requests by clinicians. This method can be used to reduce overuse of medical radiation, which poses significant health risks and staggering costs especially with regard to children.
Modular bibliometric information system with proprietary software (MOBIS-ProSoft): a versatile approach to bibliometric research tools
Sotolongo-Aguilar, Suárez-Balseiro and Guzmán-Sánchez (2000) LIBRES ISSN 1058-6768
A platform including artificial neural networking software for bibliometric research was developed to provide low complexity framework software that is reasonably available. The approach works smoothly with small and medium size corpora and may be very useful for both research and educational purposes. SOMine was used to allow work without models and statistical assumptions. The method lead to a very good representation of high-dimensional data by maintaining similarities implicit in the data.
Business intelligence
Facilitating Decision Support in Hospital Emergency Departments: A Process-Oriented Perspective
Ceglowski A., L. Churilov, J. Wassertheil (2005), European Conference on Information Systems, Regensburg, Germany
This paper outlines the identification of emergency department treatment processes and discusses how this treatment process perspective assists in framing optimization of resource utilisation, clinical decision making, training and emergency department layout.
Viscovery SOMine was used to group patients on similarity of combination and timing of clinical procedures in order to determine common clusters of primary patient treatments undertaken in the emergency department.
Customer Portfolio Analysis Using the SOM
Holmbom, A.H., T. Eklund, B. Back (2008), 19th Australasian Conference on Information Systems, Christchurch, New Zealand
The purpose of the paper is to illustrate how Self-Organizing Maps (SOMs) can be used for customer portfolio analysis (CPA), a category of CRM which serves to identify profitable customers in order to develop marketing strategies. Customer data were provided by a case company.
Viscovery SOMine was used to generate the model grouping the customers into segments with similar profiles, which were then analyzed in light of product sales information. The model could potentially be used to adjust marketing efforts to increase the profitability of customers.
Data mining trends and technologies
Die digitalen Entscheidungshelfer: Ein Markt im Umbruch (German language)
(2008) FORMAT, Edition 22/08 from 30.5.2008
Viscovery was named as an established provider of data mining solutions with superior customer relations.
Data visualization of asymmetric data using Sammon mapping and applications of self-organizing maps
Li (2005) Digital Repository at the University of Maryland, 17-Mar-2005
The performance of several software implementations of SOM-based methods were evaluated. Viscovery SOMine was found to be helpful in determining the number of clusters and recovering the cluster structure of data sets. A genocide and politicide data set is analyzed using Viscovery SOMine, followed by another analysis on the public and private college data sets with the goal to find out schools with best values.
Interactive visual interfaces: A survey
F. Murtagh, T. Taskaya, P. Contreras, J. Mothe, K. Englmeier, Artificial Intelligence Review, 19, 263-283, 2003
A new and recent implementation is reported taking concept hierarchies as input data. The visual user interfaces express domain ontologies which are based on these concept hierarchies. A web-based implementation is detailed, and examples of usage are shown. Viscovery SOMine is one of the surveyed systems.
Comparative Analysis of the Graphical Result Presentation in the SOM Software
Dzemyda, Kurasova, in Informatica 2002, Vol. 13, No. 3, 275-286, ISSN 0868-4952
Several software tools based on self-organizing maps are analyzed, using data on coastal dunes and their vegetation in Finland. The focus of the experimental comparison is the graphical result presentation. Viscovery SOMine has been reviewed in comparison to SOM-PAK, SOM-TOOLBOX, Nenet, and two academic systems.
Features, objects, and other things: ontological distinctions in the geographic domain
Mark, Skupin and Smith (2001) in Spatial Information Theory: Foundations of Geographic Information Science Lecture Notes
Response were analyzed from 263 subjects giving examples for one of five geographic categories: geographic features, geographic objects, geographic concepts, something geographic, and something that could be portrayed on a map. The frequencies of various responses were significantly different, indicating that the basic ontological terms feature, object, etc., are not interchangeable but carry different meanings when combined with adjectives indicating geographic or mappable. SOM was used for training and initial labeling of neurons, and ArcView for processing of base map configurations and two-dimensional interpolation.
A comparison of SOM neural networks and K-means clustering using real world data: Chinese consumer attitudes towards imported fruit
Sun, Collins and Kim (2001) SHS Acta Horticulturae 566: II International Symposium on Application of Modelling as an Innovative Technology in the Agri-Food Chain; MODEL-IT
SOM neural networks were compared with K-means algorithms to test their relative ability to generate reliable clustering solutions using real world data - Chinese consumer attitudes towards imported fruit. Results show that K-means performs better than SOM in terms of reliability, but SOM’s strength is its ability to discover the “natural” number of clusters.
Data mining industry: emerging trends and new opportunities
Aldana (2000) Master of Engineering in Electrical Engineering and Computer Science at the Massachusetts Institute of Technology
Viscovery SOMine is reviewed.
SOM application to qualitative information analysis
Tada, K., K. Obu-Cann (2000), ISSN:1341-9080 Faji Shisutemu Shinpojiumu Koen Ronbunshu VOL16, p.391-394
In this paper the “qualitative information analysis” is tackled, i.e. the analysis for creative definition of question and hypothesis for investigation to search for new business opportunities and potential needs from observed countless phenomenon.
Viscovery SOMine was used for the evaluation of new technology topics in several ways: to visualize the PS analysis, a planning method of the new technology topic seminar and to analyze the topics and property items of concept research.
A DACS State-of-the-Art Report - Mining Software Engineering Data: A Survey
Mendonca, M., N. L. Sunderhaft (1999), Report No. DACS-SOAR-99-3, University of Maryland
This report discusses the state-of-the-art, as well as recent advances in the use of data mining techniques as applied to software process and product information. This report includes: A discussion on data mining techniques and on how they can be used to analyze software engineering data, a bibliography on data mining with special emphasis on data mining of software engineering information, a survey of the data mining tools that are available to software engineering practitioners (including Viscovery SOMine) and a listing of web resources for data mining information.
Process optimization and engineering
Making sense of sensor data
Cook (2007) IEEE Pervasive Computing 6; 105-108
The tools and algorithms for analyzing sensor data were examined using SOMine clustering.
Design exploration of high-lift airfoil using Kriging model and data mining technique
Kanazaki, Tanaka, Jeong, and Yamamoto (2006) European Conference on Computational Fluid Dynamics, ECCOMAS CFD 2006 P. Wesseling, E. Oñate, J. Périaux (Eds) TU Delft, The Netherlands, 2006
SOMine was used for a multi-objective design exploration of a three-element airfoil consisting of a slat, a main wing, and a flap. A total of 90 sample points were evaluated using the Reynolds averaged Navier-Stokes simulation (RANS) for the construction of the Kriging model. SOM analysis was used to obtain qualitative information of the design space.
Data mining for aerodynamic design space
Jeong, Chiba and Obayashi (2005) J Aerospace Computing, Information Communication 2:452-469
SOM analysis was applied to data mining for aerodynamic design space to identify the effect of each design variable on objective functions. SOM can visualize the trade-offs among objective functions, and this information will be helpful for designers to determine the final design from non-dominated solutions of multi-objective problems. These methods were applied to two design results: a fly-back booster in reusable launch vehicle design, which has 4 objective functions and 71 design variables, and a transonic airfoil design performed with the adaptive search region method.
Visualization and data mining of pareto solutions using self-organizing map
Obayashi and Sasaki (2003)
SOMine was used to visualize tradeoffs of Pareto solutions in the objective function space for engineering design obtained by Evolutionary Computation. Based on the codebook vectors of cluster-averaged values of respective design variables obtained from the SOM, the design variable space is mapped onto another SOM. The resulting SOM generates clusters of design variables, which indicate roles of the design variables for design improvements and tradeoffs. These processes, data mining of the engineering design were applied to supersonic wing and supersonic wing-fuselage design.
Optimizing recombinant microbial fermentation processes an integrated approach
Cserjan-Puschmann, Grabherr, Striedner, Clementschitsch and Bayer (2002) BioPharm
Bioprocess automation
Bayer, Doblhoff-Dier and Mikler (2002)
SOMine was used to generate SOMs for mapping a multidimensional input matrix of online and offline data into lower dimensional output matrix sorting out the relevant relationships.
Classification of metal ions according to their complexing properties: a data-driven approach
Pletnev, Zernov in Analytica Chimica Acta, Volume 455, Issue 1, 18 March 2002, Pages 131-142
Factor, cluster and self-organizing map analyses were applied to the stability constants of complexes of metal ions and hydrogen with 3960 ligands. Both direct clustering and clustering on the basis of factor analysis established the existence of six different classes of similar cations. The self-organizing map created with Viscovery SOMine visually represents that similarity.
Application of Swarm Approach and Artificial Neural Networks for Airfoil Shape Optimization
Khurana M. S., H. Winarto, A. K. Sinha (2008), 12th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Victoria, British Columbia, Canada
The Direct Numerical Optimization (DNO) approach for airfoil shape design is presented, which requires the integration of the following modules: A geometrical shape function; Computational flow solver and; Search model for shape optimization. In this paper, the Particle Swarm Optimization (PSO) algorithm is introduced as the search agent.
Viscovery SOMine is applied to illustrate trade-offs between PSO search variables.
Airfoil Geometry Parameterization through Shape Optimizer and Computational Fluid Dynamics
Khurana, M.S., H. Winarto, A.K. Sinha (2008), 46th AIAA Aerospace Sciences Meeting and Exhibit, Nevada
This paper focuses on the design of a Re-Configurable Multi Mission Unmanned Aerial Vehicle (RC-MM-UAV) based on intelligent airfoil optimization.
Viscovery SOMine was used to analyze the effect of Yup on PARSEC Airfoil Geometry and Aerodynamics in order to optimize the shape parameters of airfoil geometry.
Application Methods for Self Organizing Map in Process Imaging for Dynamic Behavior of Aerated Agitation Vessel
Matsumoto, H., R. Masumoto, C. Kuroda (2007), Proceedings of the 10th International Congress on Engineering Applications of Neural Networks, 210-220, Thessaloniki, Greece
Viscovery SOMine is adapted to process imaging for dynamic behavior of aerated agitation vessel, and the application methods are investigated in this article. In the application, the direct imaging by CCD video camera and the PIV technology are adopted. It is shown that the generated map and clusters could give process engineers useful information about the degree of spatial dispersion of bubbles and about the determination of design parameters.
Optimized Data Exploration of Recombinant Fermentations using Neural Network Simulations
Dürrschmid, E., B. Spannbauer, G. Striedner, F. Clementschitsch, K. Bayer (1998), 7th Intern. Conference on Computer Application in Biotechnology, Osaka
In this paper a combination of SOMs and RBF networks (Radial Basis Function Networks) is presented as a powerful tool applied to fermentation data, enabling rapid recognition of interdependencies and subsequent modelling.
Viscovery SOMine was used to model the appearance and concentration of signal molecules in order to get a better understanding of the relations between metabolic load and recombinant protein production, to make use of the cell´s synthetic capacity.
Market and finance analytics
2006 General Insurance Industry Survey
Fitzgerald, McCarthy, Franks and Collins (2006) Australia Equity Research
The 2006 JP Morgan Deloitte General Insurance Survey provides a detailed overview of the current state of the Australian general insurance industry and the industry’s expectations. The report delivers critical information on the key elements of the industry from direct underwriters, reinsurers and brokers.
A comparison of supervised and unsupervised neural networks in predicting bankruptcy of Korean firms
Lee, Booth, Alam in Expert Systems with Applications, Volume 29, Issue 1, July 2005, Pages 1-16
In this study, two learning paradigms of neural networks, supervised versus unsupervised, are compared in terms of prediction accuracy in the area of bankruptcy prediction. The back-propagation (BP) network was used as representative type for supervised networks and the Kohonen self-organizing feature map (Viscovery SOMine) was used as the representative for unsupervised neural networks.
The self-organizing map in financial benchmarking
Eklund, Tomas (2004), Dissertation, Institute for Advanced Management Systems Research, Åbo Akademi University, Åbo, Finland
In this dissertation, a self-organizing map (SOM) model for financial benchmarking in the international pulp and paper industry has been built on the basis of seven financial ratios measuring different aspects of financial performance.
Viscovery SOMine was used to automatically identify the clusters at a two-level clustering using Ward’s method.
Forecasting of credit classes with the self-organizing maps
Merkevicius and Garšva (2004) INFORMACINES TECHNOLOGIJOS IR VALDYMAS 4:61-66
The use of SOMs for forecasting credit classes was examined and shown to reduce misclassification errors. Viscovery SOMine was chosen for the examination because of parameter flexibility for the learning procedure.
EUNITE - European network on intelligent technologies for smart adaptive systems
Carlsson and Kaymak (2003)
Erasmus University Rotterdam has developed a method based on SOMs and using SOMine for clustering companies based on their financial characteristics.
European financial cross-border consolidation: at the crossroads in Europe? By exception, evolution or revolution?
Abraham and Dijcke (2002) Société Universitaire Européenne de Recherches Financières
An empirical evidence of the financial performance of lean production adoption: a self-organizing neural networks approach
Biscontri and Park (2000) IEEE-INNS-ENNS International Joint Conference on Neural Networks (IJCNN'00)-Volume 5 p. 5297
A feasibility study was performed to evaluate the application of self-organizing mapping neural networks (SOMs) to examine the financial performance of the U.S. lean production firms. Using control group design, the ability of SOMs to distinguish the financial performance between members of the target group (lean firm) and control group (non-lean firm) was tested. The financial performance we investigate includes return on assets (ROA), current ratio, and the ratio of cost of goods sold to sale, gross profit ratio, asset turnover, and inventory turnover ratio. Results show that the SOMs models successfully identify the financial performance of the lean firms from non-lean firms with quantization error of less than 0.01.
An Economic Analysis of Government Transfers with Viscovery SOMine
SHIBATA, SAKAI, TAKEMITSU, MOCHIZUKI (2000) in Faji Shisutemu Shinpojiumu Koen Ronbunshu, Vol 16 p 403-404
In this paper central government financial transfer to local government is analyzed using Viscovery SOMine.
Bewertung von Investmentfonds mittels Self-Organizing Maps (German language)
Wirtz, Peter (2005), Bachelor Thesis, HfB – Business School of Finance & Management, Frankfurt
Am Beispiel des deutschen Marktes für Investmentfonds untersucht diese Arbeit, inwieweit die nach Feri TRUST klassifizierten Fonds tatsächliche Ähnlichkeiten aufweisen.
Viscovery SOMine wurde für die Bewertung von Investmentfonds benutzt.
Forecasting of Credit Classes with Self-Organizing Maps
Merkevicius E., G. Garšva, R. Simutis (2004), ISSN 1392, 124X INFORMACINES TECHNOLOGIJOS IR VALDYMAS, Nr.4(33)
In this paper the capabilities of SOMs in forecasting of credit classes are investigated.
Viscovery SOMine was used to generate a model of credit units by similar characteristics of the process in order to determine credit classes. It is shown that SOMs may distinctly reduce misclassification errors.
Credit Rating Prediction using Self-Organizing Maps
Tan, Roger P.G.H. (2000), Master thesis, Erasmus University Rotterdam
In this thesis the relationship between the financial statement of a company and its assigned credit rating is analyzed to show how much of a company’s rating is affected by the qualitative analysis performed by the rating agency. The thesis focuses on the application of SOMs to get insightful visualizations of large datasets.
Viscovery SOMine was used for credit rating predictions.
Marketing Mix Decision Making Using Scanner Data and Self-Organizing Maps
Krycha K., G. Kranner (1998), Proceedings of 2nd Slovak Conference on Artificial Neural Networks (SCANN98), Smolenice, Slovakia, ISBN 80-967935-2-7
In this contribution an application of Self-Organizing Maps is proposed to the problem of managing the product line with respect to price decisions.
Viscovery SOMine was used for the analysis of price elasticity.
Web analytics
Intelligent web traffic mining and analysis
Wanga, Abrahamb, and. Smith (2005) J Network Comp. Appl. 28:147-165
SOMine was used to generate cluster information for pattern analysis in combination with a fuzzy inference system to capture the chaotic trend to provide short-term (hourly) and long-term (daily) Web traffic trend predictions. Empirical results demonstrated that approach is efficient for mining and predicting Web server traffic and could be extended to other Web environments as well.
Improved Web searching through neural network based index generation
Wang, Alahakoon and Smith (2003) Lecture Notes in Computer Science, Springer Berlin / Heidelberg, 0302-9743 (Print) 1611-3349 (Online), Volume 2659/2003 ISBN 978-3-540-40196-4
SOMs were used for clustering query logs to identify prominent groups of user query terms for further analysis. Such groups can provide meaningful information regarding web users’ search interests. Identified clusters can further be used for developing an adaptive indexing database for improving conventional search engine efficiency. The proposed hybrid model which combines neural network and indexing for web search applications can provide better data filtering effectiveness and efficiently adapt to the changes based on the web searchers’ interests or behavior patterns.
Image and speech recognition
Using self-organizing maps for object classification in image analysis
Heiss-Czedik and Bajla (2005) Measurement Sci. Rev. 5:11
The recombinant form (rEpo) of eerythropoietin, a hormone used for doping, involves analysis of Epo chemiluminescence images containing bands. A research project funded by the World Anti-Doping Agency was established to develop software for Epo testing. The 506 objects of the training set were ordered by SOMs using Viscovery SOMine, which was chosen because of its visualization capabilities. After segmentation of band data, artifacts must be separated from the authentic bands. A classification method based on self-organizing map performs well when compared with other classification methods.
Neural networks for text-to-speech phoneme recognition
Embrechts and Arciniegas (2000) 2000 IEEE International Conference on Systems, Man, and Cybernetics, 5:3582-3587; Digital Object Identifier 10.1109/ICSMC.2000.886565
Two different artificial neural network (ANN) approaches were used for phoneme recognition for text-to-speech applications: staged backpropagation neural networks and self-organizing maps. Several current commercial approaches rely on an exhaustive dictionary approach for text-to-phoneme conversion. Applying neural networks to phoneme mapping for text-to-speech conversion creates a fast distributed recognition engine. This engine not only supports the mapping of missing words in the database, but it can also mitigate contradictions related to different pronunciations for the same word. The ANNs presented in this work were trained based on the 2,000 most common words in American English. Performance metrics for the 5,000, 7,000 and 10,000 most common words in English were also estimated to test the robustness of these neural networks.
