Interval data clustering using self-organizing maps based on adaptive Mahalanobis distances

Hajjar, Chantal; Hamdan, Hani

doi:10.1016/j.neunet.2013.04.009

Cited by 27 publications

(10 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Artificial neural network (ANN) is one of the most efficient tools that work based on artificial ℎ ℎ , * Strategies to develop robust neural network models: prediction of flash point as a case study Amin.alibakhshi@hotmail.com;aalibakhshi@geomar.de Geomar Helmholtz Center for Ocean Research Kiel, Wischhofstrasse 1-3, 24148 Kiel, Germany M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT 2 intelligence and machine learning. ANNs are capable of doing several tasks such as function approximation 1 , pattern recognition 2 , data clustering 3 , prediction of time series 4 , and so on. To provide the best performance, various types of neural networks are developed and characterized depending on the application.…”

Section: Introductionmentioning

confidence: 99%

Strategies to develop robust neural network models: Prediction of flash point as a case study

Alibakshi

2018

Analytica Chimica Acta

View full text Add to dashboard Cite

Artificial neural network (ANN) is one of the most widely used methods to develop accurate predictive models based on artificial intelligence and machine learning. In the present study, the important practical aspects of developing a reliable ANN model e.g. appropriate assignment of the number of neurons, number of hidden layers, transfer function, training algorithm, dataset division and initialization of the network are discussed. As a case study, predictability of the flash point for a dataset of 740 organic compounds using ANNs was investigated via a total number of 484220ANNs to allow covering a wide range of parameters affecting the performance of an ANN. Among all studied parameters, the number of neurons or layers was found to be the most important parameters to develop a reliable ANN with low overfitting risk. To evaluate appropriate number of neurons and layers, a value of equal or greater than 10 for the ratio of the training samples to the ANN constants was suggested as a rule of thumb. More ever, a strategy for evaluation of the authentic performance of ANNs and deciding about the reliability of an ANN model was proposed which is applicable to other models developed by supervised learning. Based on the introduced considerations, an ANN model was proposed for predicting the flash point of pure organic compounds. According to the results, the new model was found to produce the lowest error compared to other available models.

show abstract

Section: Introductionmentioning

confidence: 99%

Strategies to develop robust neural network models: Prediction of flash point as a case study

Alibakshi

2018

Analytica Chimica Acta

View full text Add to dashboard Cite

show abstract

“…In the literature, examples of fields in which interval databased information is analyzed are: pattern recognition [25], web [33], telecommunications [31], meteorology [51,71,33], environmetrics [7], chemometrics [59,[34][35][36], medicine [3,81,85], physics [47], education [22,85], economics and finance [45].…”

Section: Introductionmentioning

confidence: 99%

“…From a methodological point of view, interesting contributions on interval-valued data analysis have been proposed in principal component analysis [34,25,70], multidimensional scaling [24,47], regression analysis [76,28,79,80,45], regression trees [71,7], time series analysis [2,86], discriminant analysis [26,27], self-organizing maps [31,5,51]; cluster analysis [9,[17][18][19][20]23,50,52,63].…”

Section: Introductionmentioning

confidence: 99%

Fuzzy c -ordered medoids clustering for interval-valued data

D’Urso

Łęski

2016

Pattern Recognition

View full text Add to dashboard Cite

Fuzzy clustering for interval-valued data helps us to find natural vague boundaries in such data. The Fuzzy c-Medoids Clustering (FcMdC) method is one of the most popular clustering methods based on a partitioning around medoids approach. However, one of the greatest disadvantages of this method is its sensitivity to the presence of outliers in data. This paper introduces a new robust fuzzy clustering method named Fuzzy c-Ordered-Medoids clustering for interval-valued data (FcOMdC-ID). The Huber's M-estimators and the Yager's Ordered Weighted Averaging (OWA) operators are used in the method proposed to make it robust to outliers. The described algorithm is compared with the fuzzy c-medoids method in the experiments performed on synthetic data with different types of outliers. A real application of the FcOMdC-ID is also provided.

show abstract

“…In the literature, several statistical methods have been proposed for analyzing interval-valued data in different empirical fields, such as chemometrics [12][13][14]25], ecotoxicology [2], meteorology [24,32], medicine [35], physics [22], pattern recognition [18], telecommunications [15], health and retirement [34], economics and finance [20]. As remarked by Manski and Tamer "researchers often have only interval data on variables that can, in principle, be measured more precisely" [34].…”

Section: Introductionmentioning

confidence: 99%

“…In fact, it is more informative to consider the interval between the minimum and the maximum daily temperature than the average value or the central value of the daily temperature. From a methodological point of view, interesting contributions have been suggested in the main fields of statistics as time series analysis [3,38], cluster analysis [13,16], principal component analysis [12,18,31], multidimensional scaling [17,22], regression analysis [11,20,34], decision trees [32], self-organizing maps [2,15,24], neural networks [9].…”

Section: Introductionmentioning

confidence: 99%

Regime change analysis of interval-valued time series with an application to PM10

Cappelli

D’Urso

Iorio

2015

Chemometrics and Intelligent Laboratory Systems

View full text Add to dashboard Cite

The aim of this paper is to conduct change point analysis of interval-valued time series employing a regression trees approach. In order to deal with such time series we propose to employ a suitable distance measure that takes into account the underlying structure of interval data. Simulation results pertaining to the behavior of the proposed approach as well as an empirical application on a daily sample of air pollutant are provided, that illustrate the practical usefulness of the proposed method

show abstract

Interval data clustering using self-organizing maps based on adaptive Mahalanobis distances

Cited by 27 publications

References 6 publications

Strategies to develop robust neural network models: Prediction of flash point as a case study

Strategies to develop robust neural network models: Prediction of flash point as a case study

Fuzzy c -ordered medoids clustering for interval-valued data

Regime change analysis of interval-valued time series with an application to PM10

Contact Info

Product

Resources

About