Fig. 4From: Outbreak detection for temporal contact dataTraining and testing on different periods (Example 1). a Relative penalty reduction for the DL objective and different training periods. Testing is always done in year 3 (Y3). For example, the blue curve indicates the detection performance when node selection was done based on the network of year 1 (Y1). The dashed curve represents detection performance of the degree heuristic. b Relative penalty reduction for the DT objective and different training periods. c Relative penalty reduction for the PA objective and different training periodsBack to article page