These are the homeworks for the deep learning course.
The topics including:
-
Naive NN on MNIST dataset using stochastic gradient descent (CPU training);
-
Convolution NN with one hidden layer of Logistic Regression on MNIST dataset using stochastic gradient descent (CPU training);
-
Deep Convolution NN on CIFAR10 dataset (GPU training): features include (A) dropout, (B) trained with RMSprop or ADAM, (C) data augmentation;
-
ResNet on CIFAR100, Tiny-ImageNet datasets (GPU training): features include (A) build my own ResNet-18 for CIFAR100 and Tiny-ImageNet datasets, (B) Pretrained-ResNet18 for CIFAR100 dataset, (C) Distributed Training with both Synchronized mode and Asynchronized mode;