Measures of Diversity in Classi
نویسنده
چکیده
Diversity among the members of a team of classiiers is deemed to be a key issue in classiier combination. However, measuring diversity is not straightforward because there is no generally accepted formal deenition. We have found and studied ten statistics which can measure diversity among binary classiier outputs (correct or incorrect vote for the class label): four averaged pairwise measures (the Q statistic, the correlation, the disagreement and the double fault) and six non-pairwise measures (the entropy of the votes, the diiculty index, the Kohavi-Wolpert variance , the interrater agreement, the generalized diversity, and the coincident failure diversity). Four experiments have been designed to examine the relationship between the accuracy of the team and the measures of diversity, and among the measures themselves. Although there are proven connections between diversity and accuracy in some special cases, our results raise some doubts about the usefulness of diversity measures in building classiier ensembles in real-life pattern recognition problems.
منابع مشابه
Can Diversity amongst Learners Improve Online Object Tracking?
We present a novel analysis of the state of the art in object tracking with respect to diversity found in its main component, an ensemble classi er that is updated in an online manner. We employ established measures for diversity and performance from the rich literature on ensemble classi cation and online learning, and present a detailed evaluation of diversity and performance on benchmark seq...
متن کاملThe Relationship between Syntactic and Lexical Complexity in Speech Monologues of EFL Learners
: This study aims to explore the relationship between syntactic and lexical complexity and also the relationship between different aspects of lexical complexity. To this end, speech monologs of 35 Iranian high-intermediate learners of English on three different tasks (i.e. argumentation, description, and narration) were analyzed for correlations between one measure of sy...
متن کاملFeature selection using Fuzzy Entropy measures with Yu ' s Similarity measure
In this study, feature selection in classi cation based problems is highlighted. The role of feature selection methods is to select important features by discarding redundant and irrelevant features in the data set, we investigated this case by using fuzzy entropy measures. We developed fuzzy entropy based feature selection method using Yu's similarity and test this using similarity classi er. ...
متن کاملMeta-Evolutionary Ensembles
Ensemble methods have shown the potential to improve on the performance of individual classi ers as long as the members of the ensamble are suÆciently diverse. Individual classi ers have been trained for example on selected subsets of the records or on projections of the feature space to produce diversity. The resulting ensembles reect a priori decisions about how to allocate records or feature...
متن کامل