关键词 > 198:462
198:462 Introduction to Deep Learning Spring 2023
发布时间:2023-02-08
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
198:462 Introduction to Deep Learning
Spring 2023
Regular class time:
Lectures: Monday and Wednesday 2-3:20 pm - WL-AUD (Wright Rieman Laboratories – Chemistry Building – Busch Campus)
Recitations: Monday 4:05 -5pm - WL-AUD
Instructor Office hours: Tuesday 6-7pm (online via zoom)
https://rutgers.zoom.us/my/elgammal?pwd=TXI4ZlVPN0NYV01nV2JYbjBscjJCQT09
Class TAs/Graders: TBA
Class Web page: Canvas page
Canvas web site for the class where the assignment, announcements, grades, and other resources will be posted.
Please upload your photo to canvas so I know who is who.
Overview:
This is a basic undergraduate-level course that intends to cover a variety of fundamental deep learning topics to get you acquainted with the field.
Topics:
• Intro to Machine Learning
• From Linear Regression to Perceptron
• Multilayer Perceptron, forward and backward Propagation
• Convolution Neural Network Models
• Recurrent Neural Network, LSTMs
• Encoder-Decoder Architectures, Sequence to Sequence learning
• Attention and Transformer
• Application domains: Computer Vision and Natural Language Processing
Textbooks:
We will mainly follow:
Dive into Deep Learning by Aston Zhang, Zack Lipton, Mu Li, and Alex Smola. (https://d2l.ai/)
Other useful textbooks :
§ Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville. (https://www.deeplearningbook.org)
§ For advanced Machine Learning Topics: Pattern Recognition and Machine Learning (PRML) by Christopher C. Bishop (https://www.microsoft.com/en-us/research/uploads/prod/2006/01/Bishop-Pattern-Recognition-and-Machine-Learning- 2006.pdf)
Required Background:
Linear algebra (250) and basic probability and statistics (206).
Formal Perquisites: (01:640:250 INTRO LINEAR ALGEBRA and 01:198:112 DATA STRUCTURES and 01:198:206 INTRODUCTION TO DISCRETE STRUCTURES II and 01:198:206 INTRODUCTION TO DISCRETE STRUCTURES II )
OR
(01:640:250 INTRO LINEAR ALGEBRA and 01:198:112 DATA STRUCTURES and 01:640:477 MATHEMATICAL THEORY OF PROBABILITY and 01:640:477 MATHEMATICAL THEORY OF PROBABILITY )
OR
(01:640:250 INTRO LINEAR ALGEBRA and 01:198:112 DATA STRUCTURES and 01:960:379 BASIC PROB AND STAT and 01:960:379 BASIC PROB AND STAT )
Python programming is a must. We will mainly use Pytorch which is a python library. If you do not know python, try to learn it in the first two weeks of the class to catch up. Some helpful resources will be provided.
Course Load:
Homework Assignments (50%): 4-5 assignments, individual, involved small programming projects and other problems
Quizzes (30%): 4-6 quizzes, online or on class (TBA).
Class Project (20%): groups of 2. Will require a proposal, presentations, and a report.
Tentative Course Outline and Schedule:
Week |
Topic |
Reading |
1 |
Intro to Deep Learning |
Ch 1 |
2 |
Machine Learning Basics Preliminaries Tensors, Linear Algebra Auto Differentiation |
Ch 2.1-2.3 Ch 2.4-2.6 |
3 |
Linear Neural Network for Regression Closed Form Gradient Descent |
Ch 3 |
4 |
Linear Regression Implementation Generalization, Weight Decay |
Ch 3.6,3.7 |
5 |
Linear Neural Net for Classification Softmax regression Image Classification Example Generalization in Classification |
Ch 4 |
6 |
Multilayer Perceptron (MLP) Forward and Backward Propagation |
Ch 5.1 – 5.3 |
7 |
Numerical Stability and Initialization, Vanishing and Exploding gradients Generalization in DL, Dropouts |
Ch 5.4 – 5.6 |
8 |
Going Deeper and Larger: Layers and Modules Lazy Initialization, Custom Layers |
Ch 6 |
9 |
Convolution Neural Networks (CNNs) |
|
10 |
Modern CNNs: AlexNet, VGGNet, ResNet, Multi-branch networks |
Ch 8 |
11 |
Recurrent Neural Networks Language Models Backpropagation Through Time |
Ch 9 |
12 |
Long Short-Term Memory (LSTM) Gated Recurrent Units (GRU) |
Ch 10 |
13 |
Encoder- Decoder Architectures Sequence-to-sequence learning U-Net |
|
14 |
Attention and Transformers |
Ch 11 |
15 |
Generative Models |
|