Wyniki wyszukiwania - BazTech

1

GrDBSCAN: A granular density-based clustering algorithm

Suchy Dawid, Siminski Krzysztof

International Journal of Applied Mathematics and Computer Science

|

2023

|

Vol. 33, no. 2

297--312

EN

Density-based spatial clustering of applications with noise (DBSCAN) is a commonly known and used algorithm for data clustering. It applies a density-based approach and can produce clusters of any shape. However, it has a drawback-its worst-case computational complexity is O(n2) with regard to the number of data items n. The paper presents GrDBSCAN: a granular modification of DBSCAN with reduced complexity. The proposed GrDBSCAN first granulates data into fuzzy granules and then runs density-based clustering on the resulting granules. The complexity of GrDBSCAN is linear with regard to the input data size and higher only for the number of granules. That number is, however, a parameter of the GrDBSCAN algorithm and is (significantly) lower than that of input data items. This results in shorter clustering time than in the case of DBSCAN. The paper is accompanied by numerical experiments. The implementation of GrDBSCAN is freely available from a public repository.

2

An outlier-robust neuro-fuzzy system for classification and regression

Siminski Krzysztof

International Journal of Applied Mathematics and Computer Science

|

2021

|

Vol. 31, no. 2

303--319

EN

Real life data often suffer from non-informative objects—outliers. These are objects that are not typical in a dataset and can significantly decline the efficacy of fuzzy models. In the paper we analyse neuro-fuzzy systems robust to outliers in classification and regression tasks. We use the fuzzy c-ordered means (FCOM) clustering algorithm for scatter domain partition to identify premises of fuzzy rules. The clustering algorithm elaborates typicality of each object. Data items with low typicalities are removed from further analysis. The paper is accompanied by experiments that show the efficacy of our modified neuro-fuzzy system to identify fuzzy models robust to high ratios of outliers.

3

A Novel Cluster Ensemble Based on a Single Clustering Algorithm

Khan Tahseen, Tian Wenhong, Kadhim Mustafa R., Buyya Rajkumar

Annals of Computer Science and Information Systems

|

2021

|

Vol. 25

127--135

EN

In recent years, several cluster ensemble methods have been developed, but they still have some limitations. They often use different clustering algorithms in both stages of the clustering ensemble method, such as the ensemble generation step and the consensus function, resulting in a compatibility issues. To deal with it, we propose a novel cluster ensemble method based on an identical clustering algorithm (CEI). Experiments on real-world datasets from various sources show that CEI improves accuracy by 5% on average compared to state-of-the-art cluster ensemble methods and by 55.54% compared to AP while consuming 44.60% less execution time.

4

Combined classifier based on feature space partitioning

Woźniak M., Krawczyk B.

International Journal of Applied Mathematics and Computer Science

|

2012

|

Vol. 22, no. 4

855-866

EN

This paper presents a significant modification to the AdaSS (Adaptive Splitting and Selection) algorithm, which was developed several years ago. The method is based on the simultaneous partitioning of the feature space and an assignment of a compound classifier to each of the subsets. The original version of the algorithm uses a classifier committee and a majority voting rule to arrive at a decision. The proposed modification replaces the fairly simple fusion method with a combined classifier, which makes a decision based on a weighted combination of the discriminant functions of the individual classifiers selected for the committee. The weights mentioned above are dependent not only on the classifier identifier, but also on the class number. The proposed approach is based on the results of previous works, where it was proven that such a combined classifier method could achieve significantly better results than simple voting systems. The proposed modification was evaluated through computer experiments, carried out on diverse benchmark datasets. The results are very promising in that they show that, for most of the datasets, the proposed method outperforms similar techniques based on the clustering and selection approach.

5

A novel fuzzy c-regression model algorithm using a new error measure and particle swarm optimization

Soltani M., Chaari A., Ben Hmida F.

International Journal of Applied Mathematics and Computer Science

|

2012

|

Vol. 22, no. 3

617-628

EN

This paper presents a new algorithm for fuzzy c-regression model clustering. The proposed methodology is based on adding a second regularization term in the objective function of a Fuzzy C-Regression Model (FCRM) clustering algorithm in order to take into account noisy data. In addition, a new error measure is used in the objective function of the FCRM algorithm, replacing the one used in this type of algorithm. Then, particle swarm optimization is employed to finally tune parameters of the obtained fuzzy model. The orthogonal least squares method is used to identify the unknown parameters of the local linear model. Finally, validation results of two examples are given to demonstrate the effectiveness and practicality of the proposed algorithm.

6

Zmodyfikowany algorytm grupowania próbek węgla wokół c-centroidów

Kowal A.

Zeszyty Naukowe. Górnictwo / Politechnika Śląska

|

2002

|

z. 255

201-213

PL

W referacie przedstawiono zmodyfikowany algorytm grupowania próbek węgla wokół c-centroidów. Klasyfikacji poddawany jest zbiór 500 próbek węgla. Program tworzy żądaną liczbę klas oraz wyznacza rozmyte współczynniki przynależności próbek do klas. Optymalny podział zapewnia procedura autokorekty.

EN

In the article modified c-means clustering algorithm of coal samples is presented. A file of 500 coal samples is divided. The program creates a required number of clusters and owns the fuzzy membership coefficients for the samples. A procedure of autocorrection assures optimal classification.