Dynamic classifier ensemble using classification confidence

Li, Leijun; Zou, Bo; Hu, Qinghua; Wu, Xiangqian; Yu, Daren

doi:10.1016/j.neucom.2012.07.026

Cited by 33 publications

(17 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The method proposed by Li et al [16] assumes that base classifiers not only make a classification decision but also return a confidence score that shows their belief that their decision is correct. Dynamic ensemble selection is performed by ordering the base classifiers according to the confidence scores and fusion is performed using weighted voting.…”

Section: Dynamic Ensemble Selectionmentioning

confidence: 99%

“…Classifier selection DS, OLA, LCA, MCB, [19,7] Ensemble pruning k-NN-based DVS, KNORA, [31] Clustering-based [14] Ordering-based [16,32,10,4] Other [17] Classifier fusion DV ensemble members randomly without replacement, stopping when the probability that the predicted class will change is below a specified level. As this approach assumes homogeneous models, it aims at speeding up the classification process without significant loss in accuracy.…”

Section: Dynamic Ensemble Selectionmentioning

confidence: 99%

See 1 more Smart Citation

Dynamic ensemble pruning based on multi-label classification

2015

View full text Add to dashboard Cite

Section: Dynamic Ensemble Selectionmentioning

confidence: 99%

Section: Dynamic Ensemble Selectionmentioning

confidence: 99%

Dynamic ensemble pruning based on multi-label classification

2015

View full text Add to dashboard Cite

“…We first propose a formulation of the C-bound for ensemble methods in complex output settings. To do so, we start from complex output predictors, with the objective to build a majority vote predictor out of these (as, e.g., in Cortes et al (2014); Li et al (2013)). This new formulation makes it possible to generalize all the classification-based results of Lacasse et al (2007), Laviolette et al (2011) andGermain et al (2015).…”

Section: Introductionmentioning

confidence: 99%

Risk upper bounds for general ensemble methods with an application to multiclass classification

et al. 2017

View full text Add to dashboard Cite

This paper generalizes a pivotal result from the PAC-Bayesian literature -the C-bound-primarily designed for binary classification to the general case of ensemble methods of voters with arbitrary outputs. We provide a generic version of the C-bound, an upper bound over the risk of models expressed as a weighted majority vote that is based on the first and second statistical moments of the vote's margin. On the one hand, this bound may advantageously be applied on more complex outputs than mere binary outputs, such as multiclass labels and multilabel, and on the other hand, it allows us to consider margin relaxations. We provide a specialization of the bound to multiclass classification together with empirical evidence that the presented theoretical result is tightly bound to the risk of the majority vote classifier. We also give insights as to how the proposed bound may be of use to characterize the risk of multilabel predictors.

show abstract

“…Meanwhile experts in computer science demonstrated that using a simple classifier is not accurate enough [9]. This indicates that approaches such as those mentioned above [3,4,5,6,7,8], which address the rust incidence rate detection using simple classifiers, lack of accuracy needed for predictions.…”

Section: Introductionmentioning

confidence: 99%

An Empirical Multi-classifier for Coffee Rust Detection in Colombian Crops

Corrales

Casas

Ledezma

et al. 2015

Computational Science and Its Applications -- ICCSA 2015

View full text Add to dashboard Cite

Abstract. Rust is a disease that leads to considerable losses in the worldwide coffee industry. In Colombia, the disease was first reported in 1983 in the department of Caldas. Since then, it spread rapidly through all other coffee departments in the country. Recent research efforts focus on detection of disease incidence using computer science techniques such as supervised learning algorithms. However, a number of different authors demonstrate that results are not sufficiently accurate using a single classifier. Authors in the computer field propose alternatives for this problem, making use of techniques that combine classifier results. Nevertheless, the traditional approaches have a limited performance due to dataset absence. Therefore we proposed an empirical multiclassifier for coffee rust detection in Colombian crops.

show abstract

Dynamic classifier ensemble using classification confidence

Cited by 33 publications

References 23 publications

Dynamic ensemble pruning based on multi-label classification

Dynamic ensemble pruning based on multi-label classification

Risk upper bounds for general ensemble methods with an application to multiclass classification

An Empirical Multi-classifier for Coffee Rust Detection in Colombian Crops

Contact Info

Product

Resources

About