This project aims to train a deep neural network in PyTorch for image classification on CIFAR-10.
- Step1: Train a model with only a single fully connected layer.
- Step2: Train a model with multiple fully connected layers. - use ReLU
- Step3: Train a model with multiple fully connected layers. - not use ReLU
- Step4: Add two convolutional layers along with maxpooling layers before the fully connected layers.
- Step5: Try multiple batch sizes (1, 4, 1000) to see the effect.
- Step6: Try multiple learning rates (10, 0.1, 0.01, and 0.001) to see the effect.
- Step7: Add some data augmentation to avoid overfitting.
- Step8: Change the loss function from Cross Entropy to Mean Squared Error and see the effect.