This schedule is tentative and often updated. Please access this page regularly to have the most recent materials. Timely announcements will be made on Piazza as well.
| Lecture | Date | Description | Materials | |
|---|---|---|---|---|
| Lecture 1 | Tue Jan 15 |
Course Introduction
Course logistics Basic Numpy |
[slides]
|
|
| Lecture 2 | Thu Jan 17 |
Basics
Basic Numpy (cont'd) Basic Tensorflow |
[slides]
|
|
| Lecture 3 | Tue Jan 22 |
Basics
Execution Order Variables and Scope Variables Save and Restore |
[slides]
|
|
SUPERVISED LEARNING |
||||
| Lecture 4 | Thu Jan 24 |
Supervised Learning
Feedforward Neural Networks Activation Functions Training with Backprop |
[slides]
|
|
| LAB | Tue Jan 29 |
Compute Resources for Deep Learning
Amazon AWS and Google Colab |
[slides]
|
|
| Lecture 5 | Tue Feb 05 |
Supervised Learning (cont'd)
Regularization Augmentation Save and Restore Model Pretrained Weights |
[slides]
|
|
| Lecture 6 | Thu Feb 07 |
Convolution
Convolution Operations Case Study |
[slides]
|
|
| Lecture 7 | Tue Feb 12 |
Convolution (cont'd)
Convolution Operations Case Study |
[slides]
|
|
| Assignment 1 | Tue Feb 12 |
Due: Mar 01 | [link] | |
| Lecture 8 | Thu Feb 14 |
Convolution (cont'd)
Architectures Machines vs. Human in perception What CNN sees |
[slides]
|
|
| Lecture 9 | Tue Feb 19 |
Implementing Neural Network
Modulazation Basic Components Data Handling |
[slides]
|
|
| Lecture 10 | Thu Feb 21 |
Implementing Neural Network (cont'd)
Data Handling Inference and Graph |
[slides]
|
|
| Lecture 11 | Tue Feb 26 |
Dynamic Graphs
Traning and Evaluation DropOut |
[slides]
|
|
| Lecture 12 | Thu Feb 28 |
Dynamic Graph (cont'd)
Batch Normalization |
[slides]
|
|
RECURRENT NEURAL NETWORK |
||||
| Lecture 13 | Tue Mar 05 |
Recurrent Neural Network
Motivation Composition of Functions Recurrent Cell |
[slides]
|
|
| Lecture 14 | Tue Mar 07 |
Recurrent Neural Network (cont'd)
Connect 2 RNNs Translation Embedding layer |
[slides]
|
|
| Spring Break - No classes | Tue Mar 12 |
|||
| Spring Break - No classes | Thu Mar 14 |
|||
| Lecture 15 | Tue Mar 19 |
Recurrent Neural Network
Attention Dropout tf.while_loop |
[slides1] [slides2] |
|
| Lecture 16 | Thu Mar 21 |
Coding RNN
tf.while_loop |
[slides]
|
|
| Assignment 2 | Tue Mar 21 |
Due: Apr 10 | [link] | |
| Lecture 17 | Tue Mar 26 |
Eager Execution
Eager Mode Data Gradient Tape Save and Restore |
[slides]
|
|
| Lecture 18 | Thu Mar 28 |
Distributed Tensorflow
Managing Devices Multiple GPUs Multiple Machines |
[slides]
|
|
ADVANCED TOPICS |
||||
| Lecture 19 | Tue Apr 02 |
Unsupervised Learning
Basics Encoder-Decoder |
[slides]
|
|
| Lecture 20 | Thu Apr 04 |
VAE
Variational Inference ELBO AutoEncoder to VAE Reparameterization |
[slides]
|
|
| Lecture 21 | Tue Apr 09 |
VAE Implementation
Encoder Decoder Prior Posterior Collapse |
[slides]
|
|
| Spring Carnival - No classes | Thu Apr 11 |
|
||
| Lecture 22 | Tue Apr 16 |
Generative Adversarial Networks
Motivation from AE Minimax GAN |
[slides]
|
|
| Lecture 23 | Thu Apr 17 |
GAN Implementation
MNIST Sample Training Problems Architectures |
[slides]
|
|
| Assignment 3 | Sun Apr 20 |
Due May 09 |
[link] | |
| Lecture 24 | Tue Apr 23 |
GAN Implementation
Training Problems Tricks |
[slides]
|
|
| Lecture 25 | Thu Apr 25 |
Conditional Generation
CVAE, CGAN Image Style Transfer Text Style Transfer Cross-Domain Transfer |
[slides]
|
|
| Lecture 26 | Tue Apr 30 |
Reinforcement Learning
From Supervised to RL Model-based Q-Learning |
[slides]
|
|