I am working in R with an SVM model from the e1071 library, and my question is whether I can identify those samples that the model has misclassified and are farthest from the SVM hyperplane.