Stimulative Training of Residual Networks: A Social Psychology Perspective of Loafing Peng Y e

As shown in Fig. r1, we can see that stimulative training can always improve We further verify it on various residual networks and benchmark datasets. MobileNetV3 are single branch structure. We show the trajectory of training loss and test accuracy when applying stimulative and common training in Fig. r2. In addition, as shown in Fig. r3, the optimal balance coefficients for MobileNetV3 on CIFAR10, MobileNetV3 on CIFAR100 and ResNet50 on CIFAR100 are 5, 10 and 10 respectively. The detailed respective training settings are given as follows.