Information theoretic combination of pattern classifiers

Meynet, Julien; Thiran, Jean-Philippe

doi:10.1016/j.patcog.2010.04.013

Cited by 32 publications

(21 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Results on a daily living and a car manufacturing scenarios confirm the suitability of these measures, as it was previously shown in other pattern recognition problems [4]. Moreover, the presented mechanism outperforms other traditional measures of diversity.…”

Section: Discussionsupporting

confidence: 83%

“…Consequently, maximizing the mutual information between the estimated and the true labels I(C;Ĉ) will decrease the probability of errors [4].…”

Section: A Information Theoretic Framework Of Classificationmentioning

confidence: 99%

“…the first two terms in equation 4) [4]. Considering the case where the ensemble decision is made by majority voting they define a measure of the information theoretic accuracy of the ensemble (ITA) as the average accuracy of individual classifiers,…”

Section: Algorithm 1 It-based Creation Of Classifier Ensemblementioning

confidence: 99%

“…However, there is still no formal method for performing this selection despite several efforts to characterize these properties [3]. Recently, information theoretic (IT) measures for accuracy and diversity have been proposed as a criteria to build classifier ensembles [4], [5]. In this paper, we test these approaches on activity recognition based on body-worn sensors.…”

Section: Introductionmentioning

confidence: 99%

“…First, we formulate the classification problem from an information theoretical perspective. Then, we present the IT-based measures of accuracy and diversity proposed by Meynet and Thiran [4] as well as some approaches to exploit these measures for ensemble formation and dynamic reconfiguration. Experimental results using two datasets of activity recognition scenarios are detailed in Section III, followed by the conclusions and future work (Section IV).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Ensemble creation and reconfiguration for activity recognition: An information theoretic approach

Chavarriaga

Sagha

Millán

2011

2011 IEEE International Conference on Systems, Man, and Cybernetics

View full text Add to dashboard Cite

Abstract-Advances in sensing, portable computing devices, and wireless communication has lead to an increase in the number and variety of sensing enabled devices (e.g. smartphones or sensing garments). Pervasive computing and activity recognition systems should be able to take advantage of these sensors, even if they are not always available or appear in runtime. These sensors can be integrated into an ensemble that fuse their information to obtain the final decision. There is therefore a need for mechanisms to select which sensors should compose the ensemble, as well as techniques for dynamically reconfigure the ensemble so as to integrate new sensors.Sensors can be integrated into an ensemble where information from each of them is fused to obtain the final decisions. From the machine learning point of view, this corresponds to the combination of classifiers where measures of the accuracy and diversity of the ensemble are used to select the elements that may lead to the highest performance. Recent works propose measures of accuracy and diversity based on an information theoretical approach. In this paper we study the use of these measures for selecting ensembles in activity recognition based on body sensor networks. Besides a comparison with traditional diversity measures (e.g., Q-, κ-statistics), we also present mechanisms to exploit these measures for the dynamic reconfiguration of the ensemble and detection of changes in the network.

show abstract

Section: Discussionsupporting

confidence: 83%

“…Consequently, maximizing the mutual information between the estimated and the true labels I(C;Ĉ) will decrease the probability of errors [4].…”

Section: A Information Theoretic Framework Of Classificationmentioning

confidence: 99%

Section: Algorithm 1 It-based Creation Of Classifier Ensemblementioning

confidence: 99%