Fig. 2From: Incorporating prior knowledge induced from stochastic differential equations in the classification of stochastic observationsExpected true error, \(E\left [\protect \epsilon ^{\text {QDA}}_{\mathbf {t}_{N}}\right ]\) and \(E\left [\protect \epsilon ^{\text {OBC}}_{\mathbf {t}_{N}}\right ]\), as a function of number of training sample paths in each class and various choices of Δ μ and κ. The dashed line shows the Bayes errorBack to article page