The paper explores the relationship between diversity measures and the accuracy of classifier ensembles. It introduces ten diversity measures, including four pairwise measures (Q statistic, correlation, disagreement, double fault) and six non-pairwise measures (entropy, difficulty index, Kohavi-Wolpert variance, interrater agreement, generalized diversity, coincident failure diversity). The study investigates how these measures relate to the accuracy of the ensemble and among themselves. While some theoretical connections between diversity and accuracy exist, the results suggest that diversity measures may not always be useful in real-world pattern recognition tasks. The paper also discusses the challenges in defining and measuring diversity, and presents experiments to examine the relationships between diversity measures and ensemble accuracy. The experiments show that some diversity measures are strongly correlated with ensemble accuracy, while others are not. The study concludes that the relationship between diversity and accuracy is complex and depends on the specific context and measures used.The paper explores the relationship between diversity measures and the accuracy of classifier ensembles. It introduces ten diversity measures, including four pairwise measures (Q statistic, correlation, disagreement, double fault) and six non-pairwise measures (entropy, difficulty index, Kohavi-Wolpert variance, interrater agreement, generalized diversity, coincident failure diversity). The study investigates how these measures relate to the accuracy of the ensemble and among themselves. While some theoretical connections between diversity and accuracy exist, the results suggest that diversity measures may not always be useful in real-world pattern recognition tasks. The paper also discusses the challenges in defining and measuring diversity, and presents experiments to examine the relationships between diversity measures and ensemble accuracy. The experiments show that some diversity measures are strongly correlated with ensemble accuracy, while others are not. The study concludes that the relationship between diversity and accuracy is complex and depends on the specific context and measures used.