Balakrishnan, P. V., M. C. Cooper, V.S. Jacob, P.A. Lewis (1994). "A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering." Psychometrika 59(4): 509-525.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....such as LBG [2] k means [3] and they exhibit a topological property, allowing to analyse the ordering of centroYds. Recent literature seems to show that the VQ performances of Kohonen maps are worst than other techniques. As an example, here are some arguments recently developed: in [4], better clustering performances are shown with standard k means algorithm; in [5] the same argument is developed to justify the use of VQ algorithm with MDS (Muldi Dimensionnal Scaling) instead of Kohonen maps; in [6] the authors argue that there exists an exponent between the underlying ....

Balakrishnan P.V. A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, vol. 59, no. 4, pp. 509-525, December 1994.

....points (both being better than the other hierarchical clustering methods) The significance of this result is that the nonzero neighborhood width applied in the beginning of the SOM iteration does not improve the clustering performance of the SOM algorithm. It was also shown by Balakrishnan et al. [BCJL94], who compared the SOM algorithm to k means clustering on 108 multivariate normal clustering problems, that if the neighborhood width does not decrease to zero, the SOM algorithm performs significantly worse than the k means clustering algorithm. Evaluating the topology preservation capability of ....

P. V. Balakrishnan, M. C. Cooper, V. S. Jacob, and P. A. Lewis. A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering. Psychometrika, 59(4):509--525, 1994.

....or visualization or even how these two purposes and goals relate to each other. In a comprehensive monograph [15] SOM is said to project and visualize high dimensional data spaces . The fact that there is a relation to clustering and visualization techniques is also well known, see e.g. [1], 10] 15] 4] and [24] Theoretical analysis of SOM concentrates on issues within the method (e.g. convergence) rather than commenting on how and for what SOM should actually be used (see [7] for a survey of results) However, there is also a considerable amount of criticism formulated both ....

....SOM concentrates on issues within the method (e.g. convergence) rather than commenting on how and for what SOM should actually be used (see [7] for a survey of results) However, there is also a considerable amount of criticism formulated both in terms of empirical and theoretical comparison. In [1] as well as [30] SOM is compared to various clustering algorithms on artificial data. In [2] SOM is compared to principal component analysis and Sammon mapping on a series of artificial and real world data sets. In [10] SOM is compared to a combined method of vector quantization plus Sammon ....

[Article contains additional citation context not shown here]

Balakrishnan P.V., Cooper M.C., Jacob V.S., Lewis P.A.: A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, Vol. 59, No. 4, 509-525, 1994.

....such as LBG [2] k means [3] and they exhibit a topological property, allowing to analyse the ordering of centrods. Recent literature seems to show that the VQ performances of Kohonen maps are worst than other techniques. As an example, here are some arguments recently developed: in [4], better clustering performances are shown with standard k means algorithm; in [5] the same argument is developed to justify the use of VQ algorithm with MDS (Muldi Dimensionnal Scaling) instead of Kohonen maps; in [6] the authors argue that there exists an exponent between the underlying ....

Balakrishnan P.V. A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, vol. 59, no. 4, pp. 509-525, December 1994.

....time to preserve the spatial ordering of the input data reflected by an ordering of the cluster centroids in a one or two dimensional output space. The latter property is closely related to multidimensional scaling (MDS) in statistics. More on SOMs relation to clustering and MDS can be found in [Balakrishnan et al. 1994], Flexer 1997] Bishop et al. 1998] and [Flexer 1999] Using the Kalman coefficients for every one second segment as input vectors for the SOM neglecting their ordering in time, the authors observed eight clearly visible clusters in the output map. Even more important they were able to ....

Balakrishnan P.V., Cooper M.C., Jacob V.S., Lewis P.A.: A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, Vol. 59, No. 4, 509-525, 1994.

....generally known as self organising maps (SOM) were originally developed for biological motivation, but they are now applied in many fields. A number of research works have since proved that SOMs are useful for fast dimensionality reduction and clustering for high dimensional data (see, e.g. Balakrishnan et al. 1994; Snyder et al. 1991; Kraajvield, et al. 1995) The SOM algorithm can be compared both with cluster analysis and multidimensional scaling. Like in partitioning methods, a unit is assigned to the cluster (neuron or node) whose representative centre m j in input space, is nearest to the unit. Like ....

Balakrishnan P.V., Cooper M.C., Jacob V.S., Lewis P.A. (1994), A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, 59, 509-525.

....or even how these two purposes and goals relate to each other. In a recent comprehensive monograph [Kohonen 97] SOM is said to project and visualize high dimensional data spaces . The fact that there is a relation to clustering and visualization techniques is also well known, see e.g. Balakrishnan et al. 94] Flexer 97] Kohonen 97] Bishop et al. 98] and [Schwenker et al. 98] Theoretical analysis of SOM concentrates on issues within the method (e.g. convergence) rather than commenting on how and for what SOM should actually be used (see [Cottrell et al. 98] for a survey of results) However, ....

....on issues within the method (e.g. convergence) rather than commenting on how and for what SOM should actually be used (see [Cottrell et al. 98] for a survey of results) However, there is also a considerable amount of criticism formulated both in terms of empirical and theoretical comparison. Balakrishnan et al. 94] as well as [Waller et al. 98] compare SOM to various clustering algorithms on artificial data. Bezdek Nikhil 95] compare SOM to principal component analysis and Sammon mapping on a series of artificial and real world data sets. Flexer 97] compares SOM to a combined method of vector ....

[Article contains additional citation context not shown here]

Balakrishnan P.V., Cooper M.C., Jacob V.S., Lewis P.A.: A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, Vol. 59, No. 4, 509-525, 1994.

.... following question has to be answered: Should SOM be used for doing VQ, MDS, both at the same time or none of them Two recent comprehensive studies comparing SOM either to traditional VQ or MDS techniques separately seem to indicate that SOM is not competitive when used for either VQ or MDS: Balakrishnan et al. 94] compare SOM to traditional K means clustering on 108 multivariate normal clustering problems with known clustering solutions and show that SOM performs significantly worse in terms of data points misclassified 1 , especially with higher numbers of clusters in the data sets. Bezdek Nikhil ....

....comparison The empirical comparison was done using a 3 factorial experimental design with 3 dependent variables. The multivariate normal distributions were generated using the procedure by [Milligan Cooper 85] which since has been used for several comparisons of cluster algorithms (see e.g. Balakrishnan et al. 94] The marginal normal distributions gave internal cohesion of the clusters by warranting that more than 99 of the data lie within 3 standard deviations (oe) External isolation was defined as having the first dimension nonoverlapping by truncating the normal distributions in the first dimension ....

Balakrishnan P.V., Cooper M.C., Jacob V.S., Lewis P.A.: A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, Vol. 59, No. 4, 509-525, 1994.

....or even how these two purposes and goals relate to each other. In a recent comprehensive monograph [Kohonen 97] SOM is said to project and visualize high dimensional data spaces . The fact that there is a relation to clustering and visualization techniques is also well known, see e.g. Balakrishnan et al. 94] Flexer 97] Kohonen 97] Bishop et al. 98] and [Schwenker et al. 98] Theoretical analysis of SOM concentrates on issues within the method (e.g. convergence) rather than commenting on how and for what SOM should actually be used (see [Cottrell et al. 98] for a survey of results) However, ....

....on issues within the method (e.g. convergence) rather than commenting on how and for what SOM should actually be used (see [Cottrell et al. 98] for a survey of results) However, there is also a considerable amount of criticism formulated both in terms of empirical and theoretical comparison. Balakrishnan et al. 94] as well as [Waller et al. 98] compare SOM to various clustering algorithms on artificial data. Bezdek Nikhil 95] compare SOM to principal component analysis and Sammon mapping on a series of artificial and real world data sets. Flexer 97] compares SOM to a combined method of vector ....

[Article contains additional citation context not shown here]

Balakrishnan P.V., Cooper M.C., Jacob V.S., Lewis P.A.: A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering, Psychometrika, Vol. 59, No. 4, 509-525, 1994.

....network models can be described in terms of, and implemented by, standard statistical techniques. Ripley s work [14] 15] along the same lines presents some empirical results comparing networks trained with different algorithms with nonparametric discriminant techniques. Balakrishnan et al. [16] report comparisons of Kohonen feature maps with traditional clustering techniques such as K means. Duin [17] makes interesting observations on techniques used to compare classifiers. An area that has remained relatively unexplored in this interdisciplinary context is the use of NN techniques that ....

P.V. Balakrishnan, M.C. Cooper, V.S. Jacob, and P.A. Lewis, "A study of the classification capabilities of neural networks using unsupervised learning: A comparison with k-means clustering," Psychometrika, vol. 59, no. 4, pp. 509--525, 1994.

No context found.

Balakrishnan, P. V., M. C. Cooper, V.S. Jacob, P.A. Lewis (1994). "A study of the classification capabilities of neural networks using unsupervised learning: a comparison with k-means clustering." Psychometrika 59(4): 509-525.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC