Table 3 Analysis of error correlation between different classifiers. For each of the classifiers listed in Table 2, a list of all test images that were incorrectly classified was generated. The correlation coefficient between the incorrectly classified images was calculated for each pair of classifiers, and the minimum, maximum, mean and standard deviation of these correlations were calculated for all pairs. This process was repeated for each feature set.

From: Boosting accuracy of automated classification of fluorescence microscope images for location proteomics