Weighting Efficient Accuracy and Minimum Sensitivity for Evolving Multi-Class Classifiers

Sánchez‐Monedero, Javier; Gutiérrez, Pedro Antonio; Fernández-Navarro, Francisco; Hervás-Martı́nez, César

doi:10.1007/s11063-011-9186-9

Cited by 19 publications

(9 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The number of nodes (S) in the hidden layer in the OP-ELM algorithm is set at the beginning to 100, since this algorithm prunes the useless neurons from the hidden layer. 3,24] improves the original ELM by using a Differential Evolution (DE) algorithm [26] and determines the optimal number of hidden nodes by a cross-validation procedure. The E − ELM CV uses the evolutionary technique to optimize the input weights and the Moore-Penrose generalized inverse to analytically determine the output weights.…”

Section: Elm Algorithms Selectedmentioning

confidence: 99%

See 1 more Smart Citation

Enforcement of the principal component analysis–extreme learning machine algorithm by linear discriminant analysis

Castaño

Fernández-Navarro

Riccardi

et al. 2015

Neural Comput & Applic

Self Cite

View full text Add to dashboard Cite

In the majority of traditional Extreme Learning Machine (ELM) approaches, the parameters of the basis functions are randomly generated and don't need to be tuned while the weights connecting the hidden layer to the output layer are analytically estimated. The determination of the optimal number of basis functions to be included in the hidden layer is still an open problem. Cross-validation and heuristic approaches (constructive and destructive) are some of the methodologies used to perform this task. Recently, a deterministic algorithm based on the Principal Component Analysis (PCA) and ELM has been proposed to assess the number of basis functions according to the number of principal components necessary to explain the 90% of the variance in the data. In this work the PCA part of the PCA-ELM algorithm is joined to the Linear Discriminant Analysis (LDA) as an hybrid means to perform the pruning of the hidden nodes. This is justified by the fact that the LDA approach is outperforming the PCA one on a set of problems. Hence the idea of combining the two approaches in a LDA-PCA-ELM algorithm that is shown to be in average better than its PCA-ELM and LDA-ELM counterparts. Moreover the performance in classification and the number of basis functions selected by the algorithm, on a set of benchmark problems, have been compared and validated in the experimental section using non parametric tests against a set of existing ELM techniques.

show abstract

Section: Elm Algorithms Selectedmentioning

confidence: 99%

“…Table 5 reports the average running time (considering also the cross-validation and the test time) of the algorithms considered. All the experiments were run using a common Matlab framework proposed in [8,24]. The proposed algorithm was developed and included in the above-mentioned framework.…”

Section: Time Complexity Analysismentioning

confidence: 99%

Enforcement of the principal component analysis–extreme learning machine algorithm by linear discriminant analysis

Castaño

Fernández-Navarro

Riccardi

et al. 2015

Neural Comput & Applic

Self Cite

View full text Add to dashboard Cite

show abstract

“…Acc values range from 0 to 100 and they represent a global performance on the classification task, not being suitable for imbalanced datasets (Sánchez-Monedero et al, 2011).…”

Section: Performance Evaluation Metricsmentioning

confidence: 99%

Simultaneous modelling of rainfall occurrence and amount using a hierarchical nominal–ordinal support vector classifier

Sánchez‐Monedero

Salcedo‐Sanz

Gutiérrez

et al. 2014

Engineering Applications of Artificial Intelligence

Self Cite

View full text Add to dashboard Cite

a b s t r a c tIn this paper we propose a novel computational system for simultaneous modelling and prediction of rainfall occurrence and amount. The proposed system is based on a hierarchical system of nominalordinal support vector classifiers, the former focussed on the prediction of the rainfall occurrence, and the latter centered in the expected rainfall amount from a set of three different ordinal classes. In addition to the proposed model, we use a novel set of predictive meteorological variables, which improve the classifiers performance in this problem. We evaluate the proposed system in a real problem of rainfall forecast at Santiago de Compostela airport, Spain, showing that the system is able to obtain an accurate prediction of occurrence and rainfall amount, and we discuss the usefulness of the proposed system as part of the airport weather forecast and warning system, in order to improve airport operations.

show abstract

“…Considering the above definitions, ordinal classification is different from nominal classification in the evaluation of the classifier performance and also in the fact that the classifier should exploit the ordinal data disposition. For the former, as an example, although accuracy (Acc) has been widely used in classification tasks, it is not suitable for some type of problems, such as imbalanced datasets [43] (very different number of patterns for each class), and ordinal datasets [44]. Then, the performance metrics must consider the order of the classes so that errors between adjacent classes should be considered less important than the ones between separated classes in the ordinal scale.…”

Section: Problem Formulationmentioning

confidence: 99%

A guided data projection technique for classification of sovereign ratings: The case of European Union 27

Sánchez‐Monedero¹,

Campoy-Muñoz²,

Gutiérrez³

et al. 2014

Applied Soft Computing

Self Cite

View full text Add to dashboard Cite

Sovereign rating has had an increasing importance since the beginning of the financial crisis. However, credit rating agencies opacity has been criticised by several authors highlighting the suitability of designing more objective alternative methods. This paper tackles the sovereign credit rating classification problem within an ordinal classification perspective by employing a pairwise class distances projection to build a classification model based on standard regression techniques. In this work the ǫ-SVR is selected as the regressor tool. The quality of the projection is validated through the classification results obtained for four performance metrics when applied to Standard & Poors, Moody's and Fitch sovereign rating data of U27 countries during the period 2007-2010. This validated projection is later used for ranking visualization which might be suitable to build a decision support system.

show abstract

Weighting Efficient Accuracy and Minimum Sensitivity for Evolving Multi-Class Classifiers

Cited by 19 publications

References 21 publications

Enforcement of the principal component analysis–extreme learning machine algorithm by linear discriminant analysis

Enforcement of the principal component analysis–extreme learning machine algorithm by linear discriminant analysis

Simultaneous modelling of rainfall occurrence and amount using a hierarchical nominal–ordinal support vector classifier

A guided data projection technique for classification of sovereign ratings: The case of European Union 27

Contact Info

Product

Resources

About