Combination, Cooperation and Selection of Classifiers: A State of the Art

Gunes, Veyis; Ménard, Michel; Loonis, Pierre; Petitrenaud, Simon

doi:10.1142/s0218001403002897

Cited by 40 publications

(11 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the field Multiple Classifier Systems, the oracle has been defined as the ideal selector which always selects, for each input pattern, the classifier that provides the correct label, if any. Accordingly, some algorithms have been proposed in the literature [3].…”

Section: Dynamic Score Selection For Multimodal Systemsmentioning

confidence: 99%

Score Selection Techniques for Fingerprint Multi-modal Biometric Authentication

Giacinto

Roli

Tronci

2005

Image Analysis and Processing – ICIAP 2005

View full text Add to dashboard Cite

Fingerprints are one of the most used biometrics for automatic personal authentication. Unfortunately, it is often difficult to design fingerprint matchers exhibiting the performances required in real applications. To meet the application requirements, fusion techniques based on multiple matching algorithms, multiple fingerprints, and multiple impressions of the same fingerprint, have been investigated. However, no previous work has investigated selection strategies for biometrics. In this paper, a score selection strategy for fingerprint multi-modal authentication is proposed. For each authentication task, only one score is dynamically selected so that the genuine and the impostor users' scores distributions are mainly separated. Score selection is performed by first estimating the likelihood that the input pattern is an impostor or a genuine user. Then, the min score is selected in case of an impostor, while the max score is selected in case of a genuine user. Reported results show that the proposed selection strategy can provide better performances than those of commonly used fusion rules.

show abstract

Section: Dynamic Score Selection For Multimodal Systemsmentioning

confidence: 99%

Score Selection Techniques for Fingerprint Multi-modal Biometric Authentication

Giacinto

Roli

Tronci

2005

Image Analysis and Processing – ICIAP 2005

View full text Add to dashboard Cite

show abstract

“…A simple yet effective way to overcome this problem, is to use a novel aggregation strategy based on Dynamic Classifier Selection (DCS) [9,10], which could reduce the number of non-competent classifiers in the classification phase. This procedure analyzes the neighbourhood of the example prior to the decision step, and removes the output for those classifiers whose related class are "far enough" in the input space area.…”

Section: Introductionmentioning

confidence: 99%

Improving Pairwise Learning Classification in Fuzzy Rule Based Classification Systems Using Dynamic Classifier Selection

Fernández¹,

Galar²,

Sanz³

et al. 2015

Proceedings of the 2015 Conference of the International Fuzzy Systems Association and the European Society for Fuzzy Logic and

View full text Add to dashboard Cite

Classification based on the One-vs-One decomposition strategy has shown a high quality for addressing those problems with multiple classes, even if the learning model enables the discrimination among several concepts. The main phase of the pairwise learning is the decision process, where the outputs of the binary classifiers are combined to give a single output. Recently, it has been shown that standard decision techniques do not take into account the influence of the non-competent classifiers, i.e. those that were not trained using the class of the query example, and this can deteriorate the performance of the model. In accordance with the former, a "Dynamic Classifier Selection" for the Onevs-One approach was proposed to alleviate this issue. It basically consists of finding those classifiers whose outputs are closest to the input example, and thus remove those ones which are not related with it. In this work, we want to analyse the goodness for the former approach using a fuzzy-type baseline classifier. Experimental results show that there is in fact a significant leap in the global performance when this model is applied, both versus the standard fuzzy rule based classification system, and the One-vs-One learning approach.

show abstract

“…This may give us reason to believe that dynamic classifier selection is better than static ensemble selection. The dynamic scheme explores the use of different classifiers for different test patterns [9][10][11][12][13][14][15]. Based on the different features or different decision regions of each test pattern, a classifier is selected and assigned to the sample.…”

Section: Introductionmentioning

confidence: 99%

“…But, in fact, the two are not mutually exclusive. Hybrid methods have been shown to be useful, in that they apply the methods for different patterns [13,14]. However, we are interested in exploring another type of approach here, because we believe that ensemble selection can be dynamic as well.…”

Section: Introductionmentioning

confidence: 99%

From dynamic classifier selection to dynamic ensemble selection

Sabourin

Britto

2008

Pattern Recognition

368

227

View full text Add to dashboard Cite

In handwritten pattern recognition, the multiple classifier system has been shown to be useful for improving recognition rates. One of the most important tasks in optimizing a multiple classifier system is to select a group of adequate classifiers, known as an Ensemble of Classifiers (EoC), from a pool of classifiers. Static selection schemes select an EoC for all test patterns, and dynamic selection schemes select different classifiers for different test patterns. Nevertheless, it has been shown that traditional dynamic selection performs no better than static selection. We propose four new dynamic selection schemes which explore the properties of the oracle concept. Our results suggest that the proposed schemes, using the majority voting rule for combining classifiers, perform better than the static selection method. ᭧

show abstract

Combination, Cooperation and Selection of Classifiers: A State of the Art

Cited by 40 publications

References 42 publications

Score Selection Techniques for Fingerprint Multi-modal Biometric Authentication

Score Selection Techniques for Fingerprint Multi-modal Biometric Authentication

Improving Pairwise Learning Classification in Fuzzy Rule Based Classification Systems Using Dynamic Classifier Selection

From dynamic classifier selection to dynamic ensemble selection

Contact Info

Product

Resources

About