Density-Based Penalty Parameter Optimization on C-SVM

Figure 3

Heterogeneous distribution of the initial instances for C-SVM, Hypothesis 2: when the number of positive instances is equal to the number of negative instances, the positive instances account for a larger area.