Ensemble of Deep Learned Features for Melanoma Classification.

摘要: The aim of this work is to propose an ensemble descriptors for Melanoma Classification, whose performance has been evaluated on validation and test datasets the melanoma challenge 2018. system proposed here achieves a strong discriminative power thanks combination multiple descriptors. represents very simple yet effective way boosting trained CNNs by composing into combining scores sum rule. Several types ensembles are considered, with different CNN architectures along learning parameter sets. Moreover used as feature extractors: input image processed response particular layer (usually classification layer, but also internal layers can be employed) treated descriptor training set Support Vector Machines (SVM).