Umich EECS 498/598: Deep Learning for Computer Vision(2020 Version)

Assignments

A1

1. Pytorch Tutorial
   • Tensor Manipulation
2. K-nearest Neighbors Algorithm

A2

1. Linear Classifier
   • SVM Classifier - Forward and Backward Propagation
   • Softmax Classifier - Forward and Backward Propagation
2. Two-layer Neural Network
   • Implement Neural Network: "input - fully connected layer - ReLU - fully connected layer - softmax"

A3

1. Fully Connected Neural Network
   • Multilayer network
   • Dropout
   • Fully-connected nets with dropout
2. Convolutional Neural Network
   • Convolutional layer & Max Pooling
   • Deep convolutional networks
   • Kaiming initialization
   • Batch Normalization
   • Deep convolutional networks with Kaiming initialization and Batch Normalization

A4

1. Pytorch API of Building Neural Networks
   • Barebones PyTorch
   • PyTorch Module API
   • PyTorch Sequential API
   • Residual Networks for image classification
2. RNN & LSTM & Attention
   • Recurrent Neural Networks(RNN)
   • Long Short-term Memory(LSTM)
   • LSTM with Attention
   • Use RNN/LSTM/LSTM with Attention to predict captions of images
3. Network Visualization
   • Saliency Maps
   • Adversarial Attack
4. Style Transfer