Description
Reading Materials-2
Fundamentals of Machine LearningLinks to an external site.
This website provides a good overview of machine learning that may help refresh your memory.
Annotated History of Modern AI and Deep LearningLinks to an external site.
Prof. Jürgen SchmidhuberLinks to an external site. of KAUST AIILinks to an external site. is commonly called “father of modern AI” for his seminal work on developing the Long Short-Term Memory (LSTM). Recently he published a paper titled “Annotated History of Modern AI and Deep Learning”, providing a comprehensive and in-depth accounts of history of modern AI dominated by the artificial neural networks (ANNs) and deep learning. It is inspiring to peruse the paper and to get familiar with how the field is evolved.
Assignment #2 – Shadow Learning and Deep Learning
1. Read Jürgen Schmidhuber’s paper
2. Post your answers to the following questions on the Discussion Forum:
1. How is linear regression related to deep learning?
2. List at least three similarities between Shallow Learning and Deep Learning
3. List fundamental differences between Shallow Learning and Deep Learning
Module 2 Discussion
Please post your questions on Module 2 in this thread.
Module 3
For the first on-ground lecture, we will go over the course logistics and have a deep dive into the basic concepts of deep learning and neural networks. We will also go through the first hands-on lab with building a neural network model from scratch using Python.
Lecture1-10-21-2023_4up.pdf
Lab #1A – Deep Learning Lab Setup
Module 3 Assignment #1 Deep Learning Lab Setup
Throughout the course, you will engage in the hands-on lab activities in parallel with minds-on theoretical exploration in order to attain the real understanding and core skills in applying the deep learning technology into solving real world problems.
As a result, you will need to build a virtual deep learning lab on your own laptop. The IDE environment and the associated tech stack are as follows:
Anaconda with Python 3.7 or later versionsLinks to an external site.
Pytorch 2.0Links to an external site.
Microsoft Visual Studio Code (MS VSC)Links to an external site.
Jupiter NotebookLinks to an external site. (with MS VSC ExtensionLinks to an external site.)
GithubLinks to an external site. (with MS VSC Extension)
Please try to download them yourself first
Module 3 Discussion
Please post your questions on Module 3 in this thread.
Module 4
Reading Materials-3
Article #1: Computer Vision Primer
Computer vision is one of the most important fields where deep learning plays a significant role. The following tutorial provides a background about computer vision and lays down a foundation for solving visual imagery that related problems with deep learning.
Computer Vision PrimerLinks to an external site.
Article #2: LeNet – GradientBased Learning Applied to Document Recognition
The seminal paper “Gradient-Based Learning Applied To Document Recognition” was written by Yann LeCun, Yoshua Bengio, Leon Bottou and Patrick Haffner in 1998. It introduced LeNet, one of the first CNN architectures that popularized the idea of convolutional neural networks (CNN). You should go through the paper to fully understand the CNN.
Gradient-Based Learning Applied To Document Recognition
Module 4
Assignment #3 – Computer Vision and CNN
Read two articles and answer the following questions:
What is the purpose of Convolutional layers in a CNN?
How do Convolutional layers reduce the number of parameters in a neural network compared to traditional dense layers?
How does pooling operation help in reducing the size of the feature map in a CNN?
What is padding and why is it used in CNNs?
What is the role of activation functions in a CNN?
What are some popular architectures for CNNs, and what are they used for?
How do we address overfitting in a CNN, and what are some techniques to prevent it?
Why CNN models are better than fully-connected neural networks to solve computer vision problems?
Module 4 Discussion 1
Please post your questions on Module 4 in this thread.
Module 5
Reading Materials – Lecture #2
For the second on-ground lecture, we will go over the computer vision and the CNN model.
Lecture2-11-18-2023_4up.pdf
Module 5 Discussion
Please post your questions on Module 3 in this thread
Module 6
Reading Materials – Lecture #3
For the third on-ground lecture, we will study the the advanced sequencing model and the generative model.
Lecture3-12-09-2023_4up.pdf
Final Project – CNN Modeling using Pytorch
Final Project
This final project is intended to get you familiar with the end-to-end training and testing pipeline to develop a CNN model using PyTorch. The reference code in Jupyter Notebook file can be found here Download here.
Your final project should meet the following requirements:
Fill out your name, your student ID, and the date in the beginning
Based on the given model, make the the following changes:
The learning rate lr = 0.05
The learning rate lr = 0.0001
Compare the accuracy for both training set and test set for Case I and Case II
Email me your final Jupyter Notebook file with a name “Firstname_Lastname.ipynb”to me with a subject line “MSCS3806 Final Project”
Module 6 Discussion
Please post your questions on Module 3 in this thread
Unformatted Attachment Preview
Reading Materials-2
1. Fundamentals of Machine LearningLinks to an external site.
• This website provides a good overview of machine learning that
may help refresh your memory.
2. Annotated History of Modern AI and Deep LearningLinks to an external site.
• Prof. Jürgen SchmidhuberLinks to an external site. of KAUST
AIILinks to an external site. is commonly called “father of modern
AI” for his seminal work on developing the Long Short-Term
Memory (LSTM). Recently he published a paper titled “Annotated
History of Modern AI and Deep Learning”, providing a
comprehensive and in-depth accounts of history of modern AI
dominated by the artificial neural networks (ANNs) and deep
learning. It is inspiring to peruse the paper and to get familiar with
how the field is evolved.
Assignment #2 – Shadow Learning and Deep
Learning
1. Read Jürgen Schmidhuber’s paper
2. Post your answers to the following questions on the Discussion Forum:
1. How is linear regression related to deep learning?
2. List at least three similarities between Shallow Learning and Deep Learning
3. List fundamental differences between Shallow Learning and Deep Learning
•
Module 2 Discussion
Please post your questions on Module 2 in this thread.
Module 3
For the first on-ground lecture, we will go over the course logistics and have a deep dive
into the basic concepts of deep learning and neural networks. We will also go through
the first hands-on lab with building a neural network model from scratch using Python.
Lecture1-10-21-2023_4up.pdf
Lab #1A – Deep Learning Lab Setup
Module 3 Assignment #1 Deep Learning Lab Setup
Throughout the course, you will engage in the hands-on lab activities in parallel with
minds-on theoretical exploration in order to attain the real understanding and core skills
in applying the deep learning technology into solving real world problems.
As a result, you will need to build a virtual deep learning lab on your own laptop. The IDE
environment and the associated tech stack are as follows:
•
•
•
•
•
Anaconda with Python 3.7 or later versionsLinks to an external site.
Pytorch 2.0Links to an external site.
Microsoft Visual Studio Code (MS VSC)Links to an external site.
Jupiter NotebookLinks to an external site. (with MS VSC ExtensionLinks to an
external site.)
GithubLinks to an external site. (with MS VSC Extension)
Please try to download them yourself first
Module 3 Discussion
Please post your questions on Module 3 in this thread.
Module 4
Reading Materials-3
Article #1: Computer Vision Primer
Computer vision is one of the most important fields where deep learning plays a
significant role. The following tutorial provides a background about computer vision and
lays down a foundation for solving visual imagery that related problems with deep
learning.
•
Computer Vision PrimerLinks to an external site.
Article #2: LeNet – GradientBased
Learning Applied to Document
Recognition
The seminal paper “Gradient-Based Learning Applied To Document Recognition” was
written by Yann LeCun, Yoshua Bengio, Leon Bottou and Patrick Haffner in 1998. It
introduced LeNet, one of the first CNN architectures that popularized the idea of
convolutional neural networks (CNN). You should go through the paper to fully
understand the CNN.
•
Gradient-Based Learning Applied To Document Recognition
Module 4
Assignment #3 – Computer Vision and CNN
Read two articles and answer the following questions:
1. What is the purpose of Convolutional layers in a CNN?
2. How do Convolutional layers reduce the number of parameters in a neural
network compared to traditional dense layers?
3. How does pooling operation help in reducing the size of the feature map in a
CNN?
4. What is padding and why is it used in CNNs?
5. What is the role of activation functions in a CNN?
6. What are some popular architectures for CNNs, and what are they used for?
7. How do we address overfitting in a CNN, and what are some techniques to
prevent it?
8. Why CNN models are better than fully-connected neural networks to solve
computer vision problems?
Module 4 Discussion 1
Please post your questions on Module 4 in this thread.
Module 5
Reading Materials – Lecture #2
For the second on-ground lecture, we will go over the computer vision and the CNN
model.
Lecture2-11-18-2023_4up.pdf
Module 5 Discussion
Please post your questions on Module 3 in this thread
Module 6
Reading Materials – Lecture #3
For the third on-ground lecture, we will study the the advanced sequencing model and
the generative model.
Lecture3-12-09-2023_4up.pdf
Final Project – CNN Modeling using Pytorch
Final Project
This final project is intended to get you familiar with the end-to-end training and testing
pipeline to develop a CNN model using PyTorch. The reference code in Jupyter
Notebook file can be found here Download here.
Your final project should meet the following requirements:
1. Fill out your name, your student ID, and the date in the beginning
2. Based on the given model, make the the following changes:
I.
The learning rate lr = 0.05
II.
The learning rate lr = 0.0001
3. Compare the accuracy for both training set and test set for Case I and Case II
4. Email me your final Jupyter Notebook file with a name
“Firstname_Lastname.ipynb”to me with a subject line “MSCS3806 Final
Project”
Module 6 Discussion
Please post your questions on Module 3 in this thread
Reading Materials-2
1. Fundamentals of Machine LearningLinks to an external site.
• This website provides a good overview of machine learning that
may help refresh your memory.
2. Annotated History of Modern AI and Deep LearningLinks to an external site.
• Prof. Jürgen SchmidhuberLinks to an external site. of KAUST
AIILinks to an external site. is commonly called “father of modern
AI” for his seminal work on developing the Long Short-Term
Memory (LSTM). Recently he published a paper titled “Annotated
History of Modern AI and Deep Learning”, providing a
comprehensive and in-depth accounts of history of modern AI
dominated by the artificial neural networks (ANNs) and deep
learning. It is inspiring to peruse the paper and to get familiar with
how the field is evolved.
Assignment #2 – Shadow Learning and Deep
Learning
1. Read Jürgen Schmidhuber’s paper
2. Post your answers to the following questions on the Discussion Forum:
1. How is linear regression related to deep learning?
2. List at least three similarities between Shallow Learning and Deep Learning
3. List fundamental differences between Shallow Learning and Deep Learning
•
Module 2 Discussion
Please post your questions on Module 2 in this thread.
Module 3
For the first on-ground lecture, we will go over the course logistics and have a deep dive
into the basic concepts of deep learning and neural networks. We will also go through
the first hands-on lab with building a neural network model from scratch using Python.
Lecture1-10-21-2023_4up.pdf
Lab #1A – Deep Learning Lab Setup
Module 3 Assignment #1 Deep Learning Lab Setup
Throughout the course, you will engage in the hands-on lab activities in parallel with
minds-on theoretical exploration in order to attain the real understanding and core skills
in applying the deep learning technology into solving real world problems.
As a result, you will need to build a virtual deep learning lab on your own laptop. The IDE
environment and the associated tech stack are as follows:
•
•
•
•
•
Anaconda with Python 3.7 or later versionsLinks to an external site.
Pytorch 2.0Links to an external site.
Microsoft Visual Studio Code (MS VSC)Links to an external site.
Jupiter NotebookLinks to an external site. (with MS VSC ExtensionLinks to an
external site.)
GithubLinks to an external site. (with MS VSC Extension)
Please try to download them yourself first
Module 3 Discussion
Please post your questions on Module 3 in this thread.
Module 4
Reading Materials-3
Article #1: Computer Vision Primer
Computer vision is one of the most important fields where deep learning plays a
significant role. The following tutorial provides a background about computer vision and
lays down a foundation for solving visual imagery that related problems with deep
learning.
•
Computer Vision PrimerLinks to an external site.
Article #2: LeNet – GradientBased
Learning Applied to Document
Recognition
The seminal paper “Gradient-Based Learning Applied To Document Recognition” was
written by Yann LeCun, Yoshua Bengio, Leon Bottou and Patrick Haffner in 1998. It
introduced LeNet, one of the first CNN architectures that popularized the idea of
convolutional neural networks (CNN). You should go through the paper to fully
understand the CNN.
•
Gradient-Based Learning Applied To Document Recognition
Module 4
Assignment #3 – Computer Vision and CNN
Read two articles and answer the following questions:
1. What is the purpose of Convolutional layers in a CNN?
2. How do Convolutional layers reduce the number of parameters in a neural
network compared to traditional dense layers?
3. How does pooling operation help in reducing the size of the feature map in a
CNN?
4. What is padding and why is it used in CNNs?
5. What is the role of activation functions in a CNN?
6. What are some popular architectures for CNNs, and what are they used for?
7. How do we address overfitting in a CNN, and what are some techniques to
prevent it?
8. Why CNN models are better than fully-connected neural networks to solve
computer vision problems?
Module 4 Discussion 1
Please post your questions on Module 4 in this thread.
Module 5
Reading Materials – Lecture #2
For the second on-ground lecture, we will go over the computer vision and the CNN
model.
Lecture2-11-18-2023_4up.pdf
Module 5 Discussion
Please post your questions on Module 3 in this thread
Module 6
Reading Materials – Lecture #3
For the third on-ground lecture, we will study the the advanced sequencing model and
the generative model.
Lecture3-12-09-2023_4up.pdf
Final Project – CNN Modeling using Pytorch
Final Project
This final project is intended to get you familiar with the end-to-end training and testing
pipeline to develop a CNN model using PyTorch. The reference code in Jupyter
Notebook file can be found here Download here.
Your final project should meet the following requirements:
1. Fill out your name, your student ID, and the date in the beginning(Mengying
Li, 284218,12/17/2023)
2. Based on the given model, make the the following changes:
I.
The learning rate lr = 0.05
II.
The learning rate lr = 0.0001
3. Compare the accuracy for both training set and test set for Case I and Case II
4. Email me your final Jupyter Notebook file with a name
“Firstname_Lastname.ipynb”(Mengying_Li)to me with a subject line
“MSCS3806 Final Project”
Module 6 Discussion
Please post your questions on Module 3 in this thread
Purchase answer to see full
attachment