Automatic Clustering Algorithm Research Articles

Customer Characterization Options for Improving the Tariff Offer

This paper deals with the classification of electricity customers on the basis of their electrical behavior. Starting from an extensive field measurement-based database of customer daily load diagrams, the authors searched for the most appropriate indices or sets of indices to be used for customer classification. They propose two original measures to quantify the degree of adequacy of each index. Using the indices as distinguishing features, they adopt an automatic clustering algorithm to form customer classes. Each customer class is then represented by its load profile. They use the load profiles to study the margins left to a distribution company for fixing dedicated tariffs to each customer class. They take into account new degrees of freedom available in the competitive electricity markets, which increase flexibility in the tariff definition under imposed revenue caps. Results of a case study performed on a set of customers of a large distribution company are presented.

Multilingual phone models for vocabulary-independent speech recognition tasks

This paper presents three different methods for developing multilingual phone models for flexible speech recognition tasks. The main goal of our investigations is to find multilingual speech units that work equally well in many languages. With such a universal set it is possible to build speech recognition systems for a variety of languages. One advantage of this approach is that acoustic–phonetic parameters in a HMM-based speech recognition system can then be shared. The multilingual approach starts with the phone sets of six languages, a total of 232 language-dependent and context-independent phone models. Then, we develop three different methods to map the language-dependent models to a multilingual phone set. The first method is a direct mapping to the phone set of the International Phonetic Association (IPA). In the second approach we apply an automatic clustering algorithm for the phone models. The third method exploits the similarities of single mixture components of the language-dependent models. Like the first method the language-specific models are mapped to the IPA inventory. In the second step an agglomerative clustering is performed on the density level to find regions of similarity between the phone models of different languages. The experiments carried out with the SpeechDat(M) database, show that the third method yields almost the same recognition rate as language-dependent models. However, using this method we achieve a huge reduction of the number of densities in the multilingual system.

Vclusters

We consider the problem of delivering an effective fine-grained clustering tool to implementors and users of object-oriented database systems. This work emphasizes on-line clustering mechanisms , as contrasted with earlier work that concentrates on clustering policies (deciding which objects should be near each other). Existing on-line clustering methods can be ineffective and/or difficult to use and may lead to poor space utilization on disk and in the disk block cache, particularly for small- to medium-size groups of objects. We introduce variable-size clusters ( Vclusters ), a fine-grained object clustering architecture that can be used directly or as the target of an automatic clustering algorithm. We describe an implementation of Vclusters in the Shore OODBMS and present experimental results that show that Vclusters significantly outperform other mechanisms commonly found in object database systems (fixed-size clusters and near hints). Vclusters deliver excellent clustering and space utilization with only a modest cost for maintaining clustering during updates.

An automatic clustering algorithm suitable for use by a computer-based tool for the design, management and continuous improvement of cellular manufacturing systems

As part of the development of Celled, a tool for the design, management and continuous improvement of cellular manufacturing systems, a need was identified for a fully automatic clustering algorithm that takes the information it needs from the Cellect database, carries out clustering and returns the cell configurations back to the user for further analysis; all without user interaction. To achieve this, an algorithm called Black Box Clustering was created. This is a modified End Load Ratio algorithm aided by the use of similarity coeffi- cients that carry out rearrangement of a workstation-part matrix to obtain a block diagonal form which is marked off to identify workstation groups and part families. This paper describes the development of Black Box Clustering and demonstrates that when tested, the algorithm was shown to be effective and versatile.

A knowledge-based approach for retrieving images by content

A knowledge based approach is introduced for retrieving images by content. It supports the answering of conceptual image queries involving similar-to predicates, spatial semantic operators, and references to conceptual terms. Interested objects in the images are represented by contours segmented from images. Image content such as shapes and spatial relationships are derived from object contours according to domain specific image knowledge. A three layered model is proposed for integrating image representations, extracted image features, and image semantics. With such a model, images can be retrieved based on the features and content specified in the queries. The knowledge based query processing is based on a query relaxation technique. The image features are classified by an automatic clustering algorithm and represented by Type Abstraction Hierarchies (TAHs) for knowledge based query processing. Since the features selected for TAH generation are based on context and user profile, and the TAHs can be generated automatically by a clustering algorithm from the feature database, our proposed image retrieval approach is scalable and context sensitive. The performance of the proposed knowledge based query processing is also discussed.

An automatic and stable clustering algorithm

A new non-hierarchical and non-conventional clustering algorithm has been developed to remove the effect of the initial guess (seed points) on the performance and convergence of the “Moving Centroid Method”. The classification is also least affected on the ordering of the data presented for clustering. The most appropriate cluster centres are obtained by studying the class strings of the initial partitions, which are generated with the help of random integers. These stable cluster centres lead to the stable partitions in a unsupervised manner.

Image segmentation by a parallel, non-parametric histogram based clustering algorithm

This paper describes a totally automatic non-parametric clustering algorithm and its application to unsupervised image segmentation. The clusters are found by mode analysis of the multidimensional histogram of the considered vectors through a non-iterative peak-climbing approach. Systematic methods for automatic selection of an appropriate histogram cell size are developed and discussed. The algorithm is easily parallelizable and is simulated on a SEQUENT parallel computer. Image segmentation is performed by clustering features extracted from small local areas of the image. Segmentation of textured, color, and gray-level images are considered. Eight-dimensional random field model based features, three-dimensional RGB components, and one-dimensional gray levels are utilized for these three types of images respectively. For texture segmentation, an image plane cluster validity procedure based on region growing of the mapped back clusters in the feature space is developed. Most of the phases are also parallelized resulting in almost linear speed ups. Quite satisfactory results are obtained in all cases.

Quantification of Progressive Diabetic Macular Nonperfusion

ABSTRACT We used the IS-2000 Image Analyzer to estimate the extent of progressive diabetic macular nonperf usion in a patient by means of an automatic clustering algorithm applied to digitized fluorescein angiograms of the patient's macula taken over time. This method may provide an objective and reproducible quantification of progressive macular nonperfusion.

A modified K-means clustering algorithm for use in isolated work recognition

Studies of isolated word recognition systems have shown that a set of carefully chosen templates can be used to bring the performance of speaker-independent systems up to that of systems trained to the individual speaker. The earliest work in this area used a sophisticated set of pattern recognition algorithms in a human-interactive mode to create the set of templates (multiple patterns) for each word in the vocabulary. Not only was this procedure time consuming but it was impossible to reproduce exactly because it was highly dependent on decisions made by the experimenter. Subsequent work led to an automatic clustering procedure which, given only a set of clustering parameters, clustered patterns with the same performance as the previously developed supervised algorithms. The one drawback of the automatic procedure was that the specification of the input parameter set was found to be somewhat dependent on the vocabulary type and size of population to be clustered. Since a naive user of such a statistical clustering algorithm could not be expected, in general, to know how to choose the word clustering parameters, even this automatic clustering algorithm was not appropriate for a completely general word recognition system. It is the purpose of this paper to present a clustering algorithm based on a standard K-means approach which requires no user parameter specification. Experimental data show that this new algorithm performs as well or better than the previously used clustering techniques when tested as part of a speaker-independent isolated word recognition system.

A modified K-means clustering algorithm for use in speaker-independent isolated word recognition

Recent studies of isolated word recognition systems have shown that a set of carefully chosen templates can be used to bring the performance of speaker-independent systems up to that of systems trained to the individual speaker. The earliest work in this area used a sophisticated set of pattern recognition algorithms in a human-interactive mode to create the set of templates (multiple patterns) for each word in the vocabulary. Not only was this procedure time consuming but it was impossible to reproduce exactly, because it was highly dependent on decisions made by the experimenter. Subsequent work led to an automatic clustering procedure which, given only a set of clustering parameters, clustered tokens with the same performance as the previously developed supervised algorithms. The one drawback of the automatic procedure was that the specification of the input parameter set was found to be somewhat dependent on the vocabulary type and size of population to be clustered. Since the user of such a statistical clustering algorithm could not be expected, in general, to know how to choose the word clustering parameters, even this automatic clustering algorithm was not appropriate for a completely general word recognition system. It is the purpose of this paper to present a new clustering algorithm based on a K-means approach which requires no user parameter specification. Experimental data show that this new algorithm performs as well or better than the previously used clustering techniques when tested as part of a speaker independent isolated word recognition system.

Classification of benign and malignant breast tumors on the basis of 36 radiographic properties.

Using a semiquantitative question sheet, a radiologist estimated 36 radiographic properties upon each of 102 pathologically proven cases of benign and malignant disease of the breast. From these properties, a probability of malignancy was assigned to each case using an automatic clustering algorithm. The algorithm examined the properties of half the cases and evolved parameters with which to assign probabilities of malignancy to unseen cases. Then the validity of these parameters was tested using the other half of cases. Every mammogram used in this study had a pathologic diagnosis indicating that the surgeon was suspicious enough of malignant disease to operate. Using the properties and decision method described in the paper, the false-positive rate was lowered 45% while keeping a false-negative rate of zero in the same set of mammograms.

An Automatic Clustering Algorithm and Its Properties in High-Dimensional Spaces

An economical technique for approximating a joint N-dimensional probability density function has been described by Sebestyen and Edie [20]. The algorithm searches for clusters of points and considers each cluster as one hyperellipsoidal cell in an N-dimensional histogram. Among the advantages of this scheme are: 1) the histogram cell descriptors-location, shape, and size-can be determined adaptively from sequentially introduced data samples of known classification and, 2) the number of cells required for a good fit can usually be held to a small number. No assumptions are required about the underlying statistical structure of the data. The algorithm requires three types of "control parameters" which critically affect its performance and are dependent upon the number of dimensions. The three factors control the birth, shape, and growth rate of the cells. Guides were presented in [20] for choosing the control parameter values. These guides functioned well for spaces of 3 dimensions or less, but did not yield usable values for spaces of greater dimensionality. This paper presents heuristics which were developed to automate the selection of the control parameters. The properties of these parameters were studied as a function of dimension. Two of the control parameters were found to be linearly related to dimension. This provides a method for determining their value by extrapolation, thereby avoiding a great deal of computation.