Fig. 1

Architecture of deep neural rejection (DNR). DNR considers different network layers and learns an SVM with the RBF kernel on each of their representations. The outputs of these SVMs are then combined using another RBF SVM, which will provide prediction scores s₁,…,s_c for each class. This classifier will reject samples if the maximum score maxk=1,…,cs_k is not higher than the rejection threshold θ. This decision rule can be equivalently represented as arg maxk=0,…,cs_k(x), if we consider rejection as an additional class with s₀=θ