Dynamic selection approaches for multiple classifier systems

Dynamic selection of the best base classifier in One versus One

“…On the other hand, Dos Santos et al [10] introduce a two step DES method: in the first step, highly accurate candidates ensembles are selected; in the second step, among those ensembles, for each test sample, the ensemble with the largest confidence level is selected. In a further work Cavalin et al [7] extend the previous work and they adapt it to Dynamic Multistage Organization strategy.…”

Constructing a multi-class classifier using one-against-one approach with different binary classifiers

“…Firstly, the best algorithm is different for each dataset [14][15][16]. The best algorithm also varies from datapoint to datapoint within a dataset [17,18]. In addition, giving variety to a dataset, such as sampling, partitioning, and decomposition, also makes the best algorithm to be different [5].…”

Forest species recognition based on dynamic classifier selection and dissimilarity feature vector representation

“…The goal of dynamic selection then is to find a subset of classifiers C * (C * ⊆C) that correctly classify a given unknown pattern Q. In the literature [16], the subset C * may be composed of a single classifier [9,43] or an ensemble of classifiers [7,20]. In general, selection is performed by estimating the competence of the classifier available in the pool on local regions of the feature space.…”

“…The feature space is divided into different partitions, and the most capable classifiers for a given unknown pattern Q, are determined. Regarding the competence measures, the literature shows that they may be based on accuracy (overall local accuracy or local class accuracy) [43], probabilistic information [9], classifier behavior computed on the output profiles [7,10], and oracle information [20,22].…”