Kohonen's Self-organizing Feature Map Research Articles

Unsupervised feature extraction using neuro-fuzzy approach

The present article demonstrates a way of formulating a neuro-fuzzy approach for feature extraction under unsupervised training. A fuzzy feature evaluation index for a set of features is newly defined in terms of degree of similarity between two patterns in both the original and transformed feature spaces. A concept of flexible membership function incorporating weighted distance is introduced for computing membership values in the transformed space that is obtained by a set of linear transformation on the original space. A layered network is designed for performing the task of minimization of the evaluation index through unsupervised learning process. This extracts a set of optimum transformed features, by projecting n-dimensional original space directly to n′-dimensional ( n′<n) transformed space, along with their relative importance. The extracted features are found to provide better classification performance than the original ones for different real life data with dimensions 3, 4, 9, 18 and 34. The superiority of the method over principal component analysis network, nonlinear discriminant analysis network and Kohonen self-organizing feature map is also established.

Recognition of Impulse Fault Patterns in Transformers Using Kohonen's Self-Organizing Feature Map

Determination of the exact nature and location of faults during impulse testing of transformers is of practical importance to the manufacturer as well as designers. The presently available diagnostic techniques more or less depend on expert knowledge of the test personnel, and in many cases are not beyond ambiguity and controversy. This paper presents an artificial neural network (ANN) approach for detection and diagnosis of fault nature and fault location in oil-filled power transformers during impulse testing. This new approach relies on high discrimination power and excellent generalization ability of ANNs in the complex pattern classification problem, and overcomes the limitations of conventional expert or knowledge-based systems in this field. In the present work, the self-organizing feature map (SOFM) algorithm with Kohonen's learning has been successfully applied to the problem with good diagnostic accuracy.

Neural networks evaluating NMR data: an approach to visualize similarities and relationships of sol-gel derived inorganic-organic and organometallic hybrid polymers.

An artificial neural network (ANN)--the Kohonen Self-Organizing Feature Map (SOM)-is used to evaluate solid-state NMR spectroscopic derived data of 72 siloxane-based phosphine and organometallic functionalized hybrid polymers. The data set consists of parameters that describe their structural features and their dynamic behavior. The ANN visualizes similarities of the investigated compounds by reducing the dimension of the data set. This allows a comparison of these polymers that was not possible beforehand because of their structural diversity.

On estimation of the number of image principal colors and color reduction through self‐organized neural networks

AbstractA new technique suitable for reduction of the number of colors in a color image is presented in this article. It is based on the use of the image Principal Color Components (PCC), which consist of the image color components and additional image components extracted with the use of proper spatial features. The additional spatial features are used to enhance the quality of the final image. First, the principal colors of the image and the principal colors of each PCC are extracted. Three algorithms were developed and tested for this purpose. Using Kohonen self‐organizing feature maps (SOFM) as classifiers, the principal color components of each PCC are obtained and a look‐up table, containing the principal colors of the PCC, is constructed. The final colors are extracted from the look‐up table entries through a SOFM by setting the number of output neurons equal to the number of the principal colors obtained for the original image. To speed up the entire algorithm and reduce memory requirements, a fractal scanning subsampling technique is employed. The method is independent of the color scheme; it is applicable to any type of color images and can be easily modified to accommodate any type of spatial features. Several experimental and comparative results exhibiting the performance of the proposed technique are presented. © 2002 Wiley Periodicals, Inc. Int J Imaging Syst Technol 12, 117–127, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.10019

Adaptive color reduction

The paper proposes an algorithm for reducing the number of colors in an image. The proposed adaptive color reduction (ACR) technique achieves color reduction using a tree clustering procedure. In each node of the tree, a self-organized neural network classifier (NNC) is used which is fed by image color values and additional local spatial features. The NNC consists of a principal component analyzer (PCA) and a Kohonen self-organized feature map (SOFM) neural network (NN). The output neurons of the NNC define the color classes for each node. The final image not only has the dominant image colors, but its texture also approaches the image local characteristics used. Using the adaptive procedure and different local features for each level of the tree, the initial color classes can be split even more. For better classification, split and merging conditions are used in order to define whether color classes must be split or merged. To speed up the entire algorithm and reduce memory requirements, a fractal scanning subsampling technique is used. The method is independent of the color scheme, it is applicable to any type of color images, and it can be easily modified to accommodate any type of spatial features and any type of tree structure. Several experimental and comparative results, exhibiting the performance of the proposed technique, are presented.

Novelty-dependent learning and topological mapping

Unsupervised topological ordering, similar to Kohonen's (1982, Biological Cybernetics, 43: 59-69) self-organizing feature map, was achieved in a connectionist module for competitive learning (a CALM Map) by internally regulating the learning rate and the size of the active neighbourhood on the basis of input novelty. In this module, winner-take-all competition and the 'activity bubble' are due tograded lateral inhibition between units. It tends to separate representations as far apart as possible, which leads to interpolation abilities and an absence of catastrophic interference when the interfering set of patterns forms an interpolated set of the initial data set. More than the Kohonen maps, these maps provide an opportunity for building psychologically and neurophysiologically motivated multimodular connectionist models. As an example, the dual pathway connectionist model for fear conditioning by Armony et al. (1997, Trends in Cognitive Science, 1: 28-34) was rebuilt and extended with CALM Maps. If the detection of novelty enhances memory encoding in a canonical circuit, such as the CALM Map, this could explain the finding of large distributed networks for novelty detection (e.g. Knight and Scabini, 1998, Journal of Clinical Neurophysiology, 15: 3-13) in the brain.

Classification of rapeseed and soybean oils by use of unsupervised pattern-recognition methods and neural networks.

Unsupervised pattern-recognition methods and Kohonen neural networks have been applied to the classification of rapeseed and soybean oil samples according to their type and quality by use of chemical and physical properties (density, refractive index, saponification value, and iodine and acid numbers) and thermal properties (thermal decomposition temperatures) as variables. A multilayer feed-forward (MLF) neural network (NN) has been used to select the most important variables for accurate classification of edible oils. To accomplish this task different neural networks architectures trained by back propagation of error method, using chemical, physical, and thermal properties as inputs, were employed. The network with the best performance and the smallest root mean squared (RMS) error was chosen. The results of MLF network sensitivity analysis enabled the identification of key properties, which were again used as variables in principal components analysis (PCA), cluster analysis (CA), and in Kohonen self-organizing feature maps (SOFM) to prove their reliability.

Cluster analysis using a validated self‐organizing method: cases of problem identification

AbstractKohonen's self‐organizing feature maps (SOFM) are commonly used in cluster analysis for problem solving. However, given a set of sample data, the cluster analysis results obtained by using the standard SOFM method can vary depending on the setting of the parameters of the SOFM. To validate the cluster analysis results, information in addition to the data for the SOFM is required. This paper reports practical cases of cluster analysis using SOFM in conjunction with a measure of validation. In these cases, multivariate data of survey were used to identify problems through validated SOFM cluster analyses. Copyright © 2001 John Wiley &amp; Sons, Ltd.

Fluorescent image classification by major color histograms and a neural network

Efficient image classification of microscopic fluorescent spheres is demonstrated with a supervised backpropagation neural network (NN) that uses as inputs the major color histogram representation of the fluorescent image to be classified. Two techniques are tested for the major color search: (1) cluster mean (CM) and (2) Kohonen's self-organizing feature map (SOFM). The method is shown to have higher recognition rates than Swain and Ballard's Color Indexing by histogram intersection. Classification with SOFM-generated histograms as inputs to the classifier NN achieved the best recognition rate (90%) for cases of normal, scaled, defocused, photobleached, and combined images of AMCA (7-Amino-4-Methylcoumarin- 3-Acetic Acid) and FITC (Fluorescein Isothiocynate)-stained microspheres.

Dynamic generation of prototypes with self-organizing feature maps for classifier design

We propose a new scheme for designing a nearest-prototype classifier using Kohonen's self-organizing feature map (SOFM). The net starts with the minimum number of prototypes which is equal to the number of classes. Then on the basis of the classification performance, new prototypes are generated dynamically. The algorithm merges similar prototypes and deletes less significant prototypes. If prototypes are deleted or new prototypes appear then they are fine tuned using Kohonen's SOFM algorithm with the winner-only update strategy. This adaptation continues until the system satisfies a termination condition. The classifier has been tested with several well-known data sets and the results obtained are quite satisfactory.

Feature extraction of chromosomes from 3-D confocal microscope images.

An investigation of local energy surface detection integrated with neural network techniques for image segmentation is presented, as applied in the feature extraction of chromosomes from image datasets obtained using an experimental confocal microscope. Use of the confocal microscope enables biologists to observe dividing cells (living or preserved) within a three-dimensional (3-D) volume, that can be visualised from multiple aspects, allowing for increased structural insight. The Nomarski differential interference contrast mode used for imaging translucent specimens, such as chromosomes, produces images not suitable for volume rendering. Segmentation of the chromosomes from this data is, thus, necessary. A neural network based on competitive learning, known as Kohonen's self-organizing feature map (SOFM) was used to perform segmentation, using a collection of statistics or features defining the image. Our past investigation showed that standard features such as the localized mean and variance of pixel intensities provided reasonable extraction of objects such as mitotic chromosomes, but surface detail was only moderately resolved. In this current work, a biologically inspired feature known as local energy is investigated as an alternative image statistic based on phase congruency in the image. This, along with different combinations of other image statistics, is applied in a SOFM, producing 3-D images exhibiting vast improvement in the level of detail and clearly isolating the chromosomes from the background. Index Terms-DIC, differential interference contrast, feature extraction, feature space, image segmentation, local energy, Morlet wavelet, phase congruency, self organizing feature map, SOFM.

Some novel classifiers designed using prototypes extracted by a new scheme based on self-organizing feature map

We propose two new comprehensive schemes for designing prototype-based classifiers. The scheme addresses all major issues (number of prototypes, generation of prototypes, and utilization of the prototypes) involved in the design of a prototype-based classifier. First we use Kohonen's self-organizing feature map (SOFM) algorithm to produce a minimum number (equal to the number of classes) of initial prototypes. Then we use a dynamic prototype generation and tuning algorithm (DYNAGEN) involving merging, splitting, deleting, and retraining of the prototypes to generate an adequate number of useful prototypes. These prototypes are used to design a "1 nearest multiple prototype (1-NMP)" classifier. Though the classifier performs quite well, it cannot reasonably deal with large variation of variance among the data from different classes. To overcome this deficiency we design a "1 most similar prototype (1-MSP)" classifier. We use the prototypes generated by the SOFM-based DYNAGEN algorithm and associate with each of them a zone of influence. A norm (Euclidean)-induced similarity measure is used for this. The prototypes and their zones of influence are fine-tuned by minimizing an error function. Both classifiers are trained and tested using several data sets, and a consistent improvement in performance of the latter over the former has been observed. We also compared our classifiers with some benchmark results available in the literature.

Application of the Self-Organizing Feature Map Algorithm in Facial Image Morphing

A new facial image morphing algorithm based on the Kohonen self-organizing feature map (SOM) algorithm is proposed to generate a smooth 2D transformation that reflects anchor point correspondences. Using only a 2D face image and a small number of anchor points, we show that the proposed morphing algorithm provides a powerful mechanism for processing facial expressions.

Mapping and fuzzy classification of macromolecular images using self-organizing neural networks

In this work the effectiveness of the fuzzy kohonen clustering network (FKCN) in the unsupervised classification of electron microscopic images of biological macromolecules is studied. The algorithm combines Kohonen's self-organizing feature maps (SOFM) and Fuzzy c-means (FCM) in order to obtain a powerful clustering technique with the best properties inherited from both. Exploratory data analysis using SOFM is also presented as a step previous to final clustering. Two different data sets obtained from the G40P helicase from B. Subtilis bacteriophage SPP1 have been used for testing the proposed method, one composed of 2458 rotational power spectra of individual images and the other composed by 338 images from the same macromolecule. Results of FKCN are compared with self-organizing feature maps (SOFM) and manual classification. Experimental results prove that this new technique is suitable for working with large, high-dimensional and noisy data sets and, thus, it is proposed to be used as a classification tool in electron microscopy.

Interactive visualization and analysis of hierarchical neural projections for data mining

Dimensionality reducing mappings, often also denoted as multidimensional scaling, are the basis for multivariate data projection and visual analysis in data mining. Topology and distance preserving mapping techniques--e.g., Kohonen's self-organizing feature map (SOM) or Sammon's nonlinear mapping (NLM)--are available to achieve multivariate data projections for the following interactive visual analysis process. For large data bases, however, NLM computation becomes intractable. Also, if additional data points or data sets are to be included in the projection, a complete recomputation of the mapping is required. In general, a neural network could learn the mapping and serve for arbitrary additional data projection. However, the computational costs would also be high, and convergence is not easily achieved. In this work, a convenient hierarchical neural projection approach is introduced, where first an unsupervised neural network--e.g., an SOM--quantizes the data base, followed by fast NLM mapping of the quantized data. In the second stage of the hierarchy, an enhancement of the NLM by a recall algorithm is applied. The training and application of a second neural network, which is learning the mapping by function approximation, is quantitatively compared with this new approach. Efficient interactive visualization and analysis techniques, exploiting the achieved hierarchical neural projection for data mining, are presented.

Fast self-organizing feature map algorithm

We present an efficient approach to forming feature maps. The method involves three stages. In the first stage, we use the K-means algorithm to select N2 (i.e., the size of the feature map to be formed) cluster centers from a data set. Then a heuristic assignment strategy is employed to organize the N2 selected data points into an N x N neural array so as to form an initial feature map. If the initial map is not good enough, then it will be fine-tuned by the traditional Kohonen self-organizing feature map (SOM) algorithm under a fast cooling regime in the third stage. By our three-stage method, a topologically ordered feature map would be formed very quickly instead of requiring a huge amount of iterations to fine-tune the weights toward the density distribution of the data points, which usually happened in the conventional SOM algorithm. Three data sets are utilized to illustrate the proposed method.

Satellite sensor image classification using cascaded architecture of neural fuzzy network

Satellite sensor images usually contain many complex factors and mixed pixels, so a high classification accuracy is not easy to attain. Especially, for a nonhomogeneous region, gray values of satellite sensor images vary greatly and thus, direct statistic gray values fail to do the categorization task correctly. The goal of this paper is to develop a cascaded architecture of neural fuzzy networks with feature mapping (CNFM) to help the clustering of satellite sensor images. In the CNFM, a Kohonen's self-organizing feature map (SOFM) is used as a preprocessing layer for the reduction of feature domain, which combines original multi-spectral gray values, structural measurements from co-occurrence matrices, and spectrum features from wavelet decomposition. In addition to the benefit of dimensional reduction of feature space, Kohonen's SOFM can remove some noisy areas and prevent the following training process from being overoriented to the training patterns. The condensed measurements are then forwarded into a neural fuzzy network, which performs supervised learning for pattern classification. The proposed cascaded approach is an appropriate technique for handling the classification problem in areas that exhibit large spatial variation and interclass heterogeneity (e.g., urban-rural infringing areas). The CNFM is a general and useful structure that can give us favorable results in terms of classification accuracy and learning speed. Experimental results indicate that our structure can retain high accuracy of classification (90% in average), while the training time is substantially reduced if our system is compared to the commonly used backpropagation network.

A self-organizing neural network approach for the single AGV routing problem

In this research, a special form of Automated Guided Vehicle (AGV) routing problem is investigated. The objective is to find the shortest tour for a single, free-ranging AGV that has to carry out multiple pick and deliver (P&D) requests. This problem is an incidence of the asymmetric traveling salesman problem which is known to be NP-complete. An artificial neural network algorithm based on Kohonen's self-organizing feature maps is developed to solve the problem, and several improvements on the basic features of self-organizing maps are proposed. Performance of the algorithm is tested under various parameter settings for different P&D request patterns and problem sizes, and compared with the optimal solution and the nearest neighbor rule. Promising results are obtained in terms of solution quality and computation time.

Multithresholding of color and gray-level images through a neural network technique

One of the most frequently used methods in image processing is thresholding. This can be a highly efficient means of aiding the interpretation of images. A new technique suitable for segmenting both gray-level and color images is presented in this paper. The proposed approach is a multithresholding technique implemented by a Principal Component Analyzer (PCA) and a Kohonen Self-Organized Feature Map (SOFM) neural network. To speedup the entire multithresholding algorithm and reduce the memory requirements, a sub-sampling technique can be used. Several experimental and comparative results exhibiting the performance of the proposed technique are presented.

Fermentation seed quality analysis with self-organising neural networks

Industrial fermentation processes operate under well defined operating conditions to attempt to minimise production variability. Variability occurs for many reasons but a long held belief is that variation in the state of the seed is highly influential. In this paper a seed stage (a batch process) of an industrial antibiotic fermentation is considered and the performance of the main production fermentations is correlated with the quality of the seed using an unsupervised Kohonen self-organising feature map (SOM). It is shown that using only seed information poor performance in the final stage fermentations can be predicted. Data from industrial penicillin G fermenters is used to demonstrate the procedure. Copyright 1999 John Wiley & Sons, Inc.