Abstract-Generating ensembles from multiple individual classifiers is a usual appraoch to raise the accuracy of the decision. For decision majority voting is a popular rule. In this paper, we generalize classic majority voting by letting a further constraint to decide whether a correct or false decision is made if k correct votes is present among the total n ones. This generalization is motivated by object detection problems, where the members of the ensemble are image processing algorithms giving their votes as pixels in the image domain. The shape of the desired object define a geometric constraint the votes should obey to be able to decide together. Namely, the votes in this scenarion should fall inside a region matching the shape of the object. We give several theoretical result in this new model for both dependent/indipendent classifiers, whose individual accuracies may also differ. As a real world example we present our ensemble-based system developed for the detection of the optic disc in retinal images. For this problem experimental results are shown on how our model is capable to characterize such a system and how the model can give a helping hand on the further improvability of the system, as well.

An object splitting model using higher-order active contours for single-cell segmentation József Molnár, Csaba Molnar, Peter Horvath #49 Változó méretű anatómiai régiók szegmentálása MRI felvételeken Tóth Márton, László Ruskó, Balázs Csébfalvi #52 Szubretinális folyadéktér és ciszta automatikus detektálása retinafelvételeken Melinda Katona, László Nyúl #10 Let the data speak! -Exploration and discovery methods for phenotypic image analysis

No abstract

No abstract

Ensemble-based approaches are very effective in various fields in raising the accuracy of its individual members, when some voting rule is applied for aggregating the individual decisions. In this paper, we investigate how to find and characterize the ensembles having the highest accuracy if the total cost of the ensemble members is bounded. This question leads to Knapsack problem with non-linear and non-separable objective function in binary and multiclass classification if the majority voting is chosen for the aggregation. As the conventional solving methods cannot be applied for this task, a novel stochastic approach was introduced in the binary case where the energy function is discussed as the joint probability function of the member accuracy. We show some theoretical results with respect to the expected ensemble accuracy and its variance in the multiclass classification problem which can help us to solve the Knapsack problem.