Figure 1From: Improving peptide-MHC class I binding prediction for unbalanced datasetsTheoretical relation between λ 2 and EK ( θ ). Theoretical relationship between the training false negative cost (λ2) that minimizes the expected cost of a classifier (EK(θ)) for a given type 2 error cost (κ2). The dotted lines represent one standard deviation from the mean. Here κ1 = 1, λ1 = 1 and π = 0.5.Back to article page