“…Considering nonlinear performance functions in SRA-RI, the kernel function for the nonlinear classification should be chosen. Due to its excellent applicability as well as high flexibility shown in many previous researches (Alibrandi, Alani, & Ricciardi, 2015;Bourinet et al, 2011;Pan & Dias, 2017;Song et al, 2013), the Gaussian kernel is employed in this work. The parameter in Gaussian kernel is defined as the mean of the pairwise distances between two types of training samples (Jaakkola, Diekhans, & Haussler, 1999).…”