Fig. 5

Total number of errors (FP + FN) as a function of the true positives (TP) in the test set. Our DL models (OursDLM) are shown in blue. In red are our traditional ML models (with optimization and variable selection). The traditional ML models without optimization and with variable selection are in orange. In purple are the traditional ML models without optimization and with no variable selection. The results of the HSI models in our database are in green.