AdaCN: An Adaptive Cubic Newton Method for Nonconvex Stochastic Optimization

Figure 3

The curves of train and test accuracy on CIFAR10. (a) Train accuracy of VGG11 on CIFAR10. (b) Test accuracy of VGG11 on CIFAR10. (c) Train accuracy of ResNet34 on CIFAR10. (d) Test accuracy of ResNet34 on CIFAR10. (e) Train accuracy of DenseNet121 on CIFAR10. (f) Test accuracy of DenseNet121 on CIFAR10.

(a)

(b)

(c)

(d)

(e)

(f)