Embedded Files
This course will cover some selected topics in deep learning, including computer vision and reinforcement learning. It will be assumed that the students have sufficient knowledge in either of TensorFlow or PyTorch. One thing to make sure is that it is impossible to cover the broad aspects of deep learning within a semester by myself. Hence, the selected topics in this course is very much biased towards what I have touched (like the blind man on right).
Lecture Notes
Lecture Notes
Syllabus
Syllabus
Week 1 - Introduction
Course Logistics
Introduction to deep learning
Historical Review
Intro to LaTeX
Week 2 - Neural Networks
Linear model
Multi layer perceptron
TF: Linear regression
Week 3 - Optimization
SGD / Adadelta / Adagrad / RMSProp / Adam
Regularization methods
TF: MLP
Week 4 - Convolutional Neural Networks (CNNs)
Basics of convolutions
AlexNet / VGG / GoogLeNet / ResNet
TF: CNN
Week 5 - Modern CNNs
Detection
Semantic segmentation
VQA
TF: Custom forward/backward paths
Week 6 - Sequential Models
RNN / LSTM / GRU
TF: Model save and restore
Week 7 - Attention Models
NMT
Transformer
BERT
TF: Ray usage + TF agent
Week 8 Project Proposal
1 page report
5 minute presentation
Week 9 - Review
Summary
Week 10 - Bayesian Neural Network
Basic probability theory
Variational inference
TF: MDN (+uncertainty estimation)
Week 11 - Generative Models
Generative adversarial networks
Variational auto-encoders
Normalizing flow
TF: GAN / VAE
Week 12 - Markov Decision Processes
Bellman equation
MDP
Value iteration / policy iteration
TF: PPO / SAC
Week 13 - Reinforcement Learning
Policy gradient
Population based methods
TF: PPO-Ray, SAC-Ray, ARS-Ray
Week 14 - Review
Summary
Week 15 - Final Project Presentation (1)
4 pages short paper
10 minute presentation + demo
Week 16 - Final Project Presentation (2)
4 pages short paper
10 minute presentation + demo
TensorFlow practice: https://github.com/sjchoi86/tf_practice
Page updated
Google Sites
Report abuse