01/08/2023

ONLINE MASTER OF SCIENCE IN ANALYTICS

OMSA 6740 – COMPUTATIONAL DATA ANALYSIS / MACHINE LEARNING I
TENTATIVE SYLLABUS SPRING 2023
H. Milton Stewart School of Industrial and Systems Engineering
Georgia Institute of Technology

PROFESSOR: Yao Xie; [email protected]

Professor Office Hour: Thursday 9:30-10:00 pm.

TEACHING ASSISTANTS:

(Head TA) Gregory Loshkajian [email protected]

Neepa Biswas [email protected]
Pengfei (Alfie) Chen [email protected]
Vishal Hariharan [email protected]
Samuel Hood [email protected]
Mark Patrick [email protected]
Sameer Reddy [email protected]
Brian Rivers [email protected]
Brent Speelman [email protected]
Cody Westgard [email protected]

PREREQUISITES
• Undergraduate level probability, linear algebra, and statistics.
• Basic programming using MATLAB or Python.

COURSE DESCRIPTION

Machine learning is a field of computer science that gives computers the ability to learn without being
explicitly programmed. The course is designed to answer the most fundamental questions about
machine learning: What are the most important methods to know about, and why? How can we answer
the questions such as “is this method better than that one' using asymptotic theory”? How can we
answer the question 'is this method better than that one' for a specific dataset of interest? What can we
say about the errors our method will make on future data? What's the “right” objective function? How
to tune parameters? What does it mean to be statistically rigorous?

This course is designed to give graduate students a thorough grounding in the methods, theory,
mathematics and algorithms needed to do research and applications in machine learning. The course
covers topics from machine learning, classical statistics, and data mining. Students entering the class
with a pre-existing working knowledge of probability, statistics and algorithms will be at an advantage,
but the class has been designed so that anyone with a strong numerate background can catch up and
fully participate. Some experience with coding is expected (in Python, or MATLAB).

For detailed course topics, please see the tentative course schedule.

LEARNING OBJECTIVES
After taking this course, students should be able to:

1
01/08/2023
• Gain thorough understanding in the methods, theory, mathematics and algorithms needed to do
research and applications in machine learning.
• Implementing and use machine learning algorithms.
• Gain experience with analyzing real data.

TEXTBOOKS/READINGS
• Textbook: The course material will be based on lectures slides provided in the course.
• Recommended References:
o (PRML) Pattern recognition and Machine Learning, Christopher M. Bishop.
o (ESL) The elements of Statistical Learning: Data Mining, Inference, and Predictions,
2nd edition, Trevor Hastie, Robert Tibshirani, and Jerome Friedman.
o (FML) Foundations of Machine Learning, 2nd edition. Mehryar Mohri, Afshin
Rostamizadeh, and Ameet Talwalkar.

COMMUNICATION
Instructor Communication: All communication from your instructor will take place in Canvas. You
are expected to check Canvas every day for important course-related information. However, by
following the course instructions, you can also ensure that you do not miss important details,
announcements, etc. by adjusting your account settings to receive important information directly to
your email account. For more details, log into Canvas, enter the course, and see the section entitled
“Before You Begin: Instructions for Getting Started.”

Communication with TAs: Each student will be assigned to one TA, who will be responsible for
grading your homework, and answering specific questions and requests. After the first week of the
semester, we will make these assignments, and provide guidance on how to find your assigned TA on
Canvas. For all your administrative requests, such as questions about grading/regrading homework
and asking for extensions, please email your assigned TA and cc the head TA. Do not message
your TA on Canvas, we cannot guarantee a response there. Your assigned TA will handle all
regrading requests.

Content Questions and Help: Because an answer to an individual’s question can often help others,
please do not email your questions directly to the instructor. Instead, course and content questions will
be addressed on Ed Discussion. Feel free to make a private post (sent to all instructors, so the
professor and all the TAs will see your question) if your question involves private material such as
your derivation work for a homework problem. For more information, log into Canvas, enter the
course, and see the section entitled “Before You Begin: Instructions for Getting Started.” Every day,
there will be at least one TA on-duty, to answer all that day’s questions on ed. Please allow some time
for TAs to respond, especially over the weekend.

Ed Discussion Post Protocols: Please be courteous when posting on Ed Discussion and treat fellow
students, the TAs, and professor with respect. In any public posts, please do not show any of your
answers related to the homework problems, such as code snippets, mathematical derivations/proofs,
etc. If you would like to show any plots (which do not disclose any explicit answer to the questions)
from your implementation in the discussion, please either make them private (only share with
instructors) and/or add watermarks to those images/results. Please make your questions specific, do
not ask, for example, for instructors to verify your results, or debug your code.

When asking questions on Ed Discussion, please be specific. In principle, instructors are not
responsible for debugging programs, and will not comment on purely coding questions. For example,
please do not send a code file to a TA, or post a question asking why a section of code “doesn't work”.
2
01/08/2023

Office Hours. Live office hours will be conducted every week via Zoom. Each TA will host their own
office hours session, and where possible, the professor will do so as well. These sessions will be both
an opportunity for the instructor to discuss course logistics and content and an opportunity for you to
ask questions. While it is strongly suggested that you participate in these meetings, all sessions will be
recorded and archived if you are unable to attend or wish to reference them later. For the meeting
schedule, links, and archives, please see the section in your Canvas course entitled “Weekly
Videoconferences.” You may attend any of the office hours that fit into your schedule.

STUDENT EFFORT
Students are expected to devote 8-10 hours per week to complete the course requirements. This
guideline encompasses all class activities, including reading the textbook and supplementary resources,
watching lesson videos, participating in office hours and forum discussions, completing homework
assignments and projects. Of course, students can spend as much time as necessary, but it is important
to be careful not to fall behind.

GRADING
Grades will be assigned on the following basis:

Homework: 70%
Project: 30%
- (Project proposal) 5%
- (Project report) 25%

Important: Make sure the scores in Canvas are consistent with the feedback you receive. We will not
make any change in grading if asked more than 2 weeks after releasing grades. We cannot accept any
homework regrading requests after the project due date. If you have any questions about your
grade/feedback, please do not respond/discuss through the Canvas Grade pages, instead, please
send an email directly to your grading TA with your questions, as well as the regrade request.

The following grading scale will be used in the course:

• 90-100%: A
• 80-89%: B
• 70-79%: C
• 60-69%: D
• below 60%: F

LESSONS
Video lessons for this class can be found on Canvas, under Media Gallery. The video lectures will be
posted according to the Course Schedule.

HOMEWORK
Due Dates and Submissions: Homework should be submitted in Canvas by 11:59 pm EST on the
date it is due. No submission will be accepted through email. We strongly encourage the use of
LaTeX for your submission. Assignments will include both exercises and programming problems; the
programming problems will ask you to carry out statistical analysis using statistical software. Keep in
mind that you should NOT hand in raw program output. Conclusions and interpretation of results
are more important than printouts. Compute output without proper explanation will not receive full

3
01/08/2023
grades. That said, notebook style submissions, with code and explanation blocks, are acceptable.
Please use MATLAB or Python for homework programming.

Please double check your homework submission to make sure it is complete (include all your answers)
before submission. Before the deadline, you can submit as many times as you like (your grader will
only review the last submission). However, we cannot accept any additional submission (nor replacing
any part of the original submission) past the deadline, without applying an appropriate penalty.

(Important) Homework code submission requirement

Please make sure you read through the below guide for homework submission and strictly follow the
instruction. For each assignment, two files need to be submitted through Canvas before the deadline:

(1) A pdf with a name: ‘Lastname_Firstname_HWx_report’. In this report file, you need to
summarize ALL your write-up answers to questions, including analytical answers, program
output/images/summaries, etc. (Please do not include your code in the pdf, unless Jupyter
Notebook is used, then you should print it into the required pdf). For this report, we strongly
recommend typing your answers. You can use either Jupyter Notebook, Latex, or any other
mark-down tool. If you choose to handwrite your report, please make sure your writing is
legible enough for your grader.
(2) A zip file with name “Lastname_Firstname_HWx.zip”: In this zip file, you need to have a
folder with the same name: ‘Lastname_Firstname_HWx.’, which should include all your
program files.

Please make sure you strictly follow the above file format requirements to avoid unnecessary loss of
points. In addition, for programming problems, you can use either Matlab or Python (3.X). When you
submit code, please also include data in the folder. Your TA will not be allowed to modify your
code, so please make sure your code doesn’t have directory dependence on your computer. In other
words, avoid the use of absolute paths. If your code has a pathing issue, it will be considered as “not
executable,” and receive no credit.

You can work together with other students on homework, as long as your write-up and solutions
represent your own work. You are also allowed (and encouraged) to ask the instruction team questions,
but keep in mind that a good faith effort to work through the material beforehand will often help you
form good questions. Any academic misconduct is subject to an F grade as well as referral to the Dean
of students.

Homework accommodations: We have the following policies to help with emergent situations:

(1) You can have up to 10 days of homework extension without penalty. Please email and notify
your assigned TA to use an extension, and make sure to cc the Head TA when emailing your
TA. Homework extension requests should be made before the original homework due date; if
you do not do this, any submission after the due date will be considered late, and be discounted
accordingly. You will not need to do anything else after asking for an extension, when grading,
your TA will keep track of the amount of time you used, and grade your assignment
accordingly. Note that you may only use 7 extension days on any single assignment. No
extension can be used for any final project deliverable.
(2) If you have already used the above 10 days of homework extension, and if you submit the
homework late: one day late the grade will be discounted to 75% of your total, two days late
the grade will discount to 50% of your total, three days late the grade will discount to 25% of
your total. Past three days, your homework will not be accepted, and you will receive no credit.
4
01/08/2023

FORMAL EXCUSES
Officially, there are 2 channels via which a student can get formal excuses.

(1) A student can request a formal exemption from the Dean's office for illness or other personal
issues by filling out a form at https://gatech-
advocate.symplicity.com/care_report/index.php/pid122802?
(2) A student may also make requests to the registrar for an institute absence, such as attending
conference, religious observances (within the first 2 weeks of the
course) https://registrar.gatech.edu/info/institute-approved-absence-form-for-students.

PROJECT

Project Team: Projects will be done in groups of 1-3 students. Please plan for the project early; you
need to form your own team (e.g., you can ask on Ed Discussion). You will also need to come up with
the topic of your project. You can discuss with professor or TAs if you need some suggestions coming
up with a topic. If you decide to do a team project, each team member will receive the same grade.
Also, your project grading will take into account the size of your team. You are responsible for the
collaboration and work division inside your group. With this in mind, the instruction team will not
provide judgement for any disputes amongst team members; such cases will be escalated to the Dean’s
office.

Project Submission: In the middle of the semester, each team will need to submit a project proposal
(see Course Schedule for the due date). By the end of the semester (see Course Schedule for the due
date), each group needs to submit one project report. No data or code submission will be required (only
proposal and report pdfs). Please see the Project Guidelines document on Canvas for additional details
and format requirements.

Project Grading:

• 5% Project Proposal (this is mainly a format and progress check; while you will receive no
detailed feedback on the proposal, you are welcome to discuss topics and ideas with the TAs
and professor).
• 35% Project Final Report (5% on creativity and novelty; 10% on formulation; 15% on the rigor
of implementation, 5% on report writing quality).

Project due dates:

Please note that the final project due date is firm, and there is no extension policy for any project
deliverable. In both cases, this is to ensure that we are able to provide feedback/grades in a timely
manner, particularly for the final report. The final project report due date is close to the university
registrar's due date for a letter grade, and we need the time to grade the final project and calculate final
letter grades. If you foresee any situation that prevents you from submitting final project deliverables
by the due date, please let us know as soon as possible. (For documented reasons, we may be able to
assign your letter grade as "incomplete" without the final project report and update the letter grade in
the next semester).

PLAGIARISM
Plagiarism is considered a serious offense. You are not allowed to copy and paste or submit materials
created or published by others, as if you created the materials. All materials submitted and posted
must be your own original work.
5
01/08/2023

STUDENT HONOR CODE

You are responsible for completing your own work. All OMS Analytics degree students are expected
and required to abide by the letter and the spirit of the Georgia Tech Honor Code. The teaching
assistants and I will also abide by the Honor Code. I am very serious about this expectation because
ethical behavior is extremely important in all facets of life. To review the Georgia Tech Honor Code,
please visit http://osi.gatech.edu/content/honor-code . Any OMS Analytics degree student suspected of
behavior in violation of the Georgia Tech Honor Code will be referred to Georgia Tech’s Office of
Student Integrity.

ACCOMMODATIONS FOR STUDENTS WITH DISABILITIES

If you are a student with learning needs that require special accommodation, contact the Office of
Disability Services at (404) 894-2563 or http://disabilityservices.gatech.edu/, as soon as possible, to
make an appointment to discuss your special needs and to obtain an accommodations letter. Please also
e-mail me as soon as possible in order to set up a time to discuss your learning needs.

Course Schedule
All assignments are due by 11:59 pm EST (electronic submission via Canvas) on the date listed below.
Week/Dates Module/Topic Weekly Overview Deliverables
Week 1 Introduction and Overview, Overview of the topics and scope of the Homework 1, release
01/09-01/15 Clustering and k-means class. We will introduce a building block 01/12, Due 01/25
of a fundamental problem in unsupervised
learning (clustering): K-means.
Week 2 Spectral Clustering We will discuss another type of clustering
01/16-01/22 algorithm: spectral clustering, which is
(01/16, MLK) different from k-means since it is based on
geometry (connectivity) of data
Week 3 Dimensionality Reduction and We will present linear dimensionality Homework 2, release
01/23-01/29 PCA reduction technique called PCA, and non- 01/26, Due 02/08
linear dimensionality reduction techniques
Week 4 Nonlinear Dimensionality Discuss basic density estimation method,
01/30-02/05 Reduction which captures the distributional
information of the data
Week 5 Density Estimation We will present a popular type of model Homework 3, release
02/06-02/12 for densities called Gaussian mixture 02/09, Due 02/22
models and discuss how to fit such models
Week 6 Gaussian Mixture Model and We will introduction the essentials of
02/13-02/19 EM Algorithm optimization theory which is a foundation
of developing machine learning algorithms
Week 7 Basic of Optimization Theory, Introduce classification problem and basic Homework 4, release
02/20-02/26 Classification Naïve Bayes and methods for classification. 02/23, Due 03/08
Logistic Regression
Week 8 Support Vector Machine Introduce SVM classifier and related
02/27-03/05 (SVM) theory.

Week 9 Neural Networks Understand basic neural networks. Homework 5, release

03/06-03/12 03/09, Due 03/25

Week 10 Feature selection and anomaly Feature selection methods and introduction
03/13-03/19 detection to anomaly detection

Week 11&12 Boosting Algorithms and Introduce basic boosting algorithms and Homework 6, release -
03/20-04/02 AdaBoost AdaBoost. 03/26, Due 04/19

6
01/08/2023
(03/20-03/24,
Spring Break)
Week 13 Random Forest Introduce tree-based methods for Project Proposal Due
04/03-04/09 regression and classification, and random 04/05.
forest
Week 14 Bias-Variance Tradeoff and Introduce principle of bias-variance
04/10-04/16 Cross-Validation tradeoff and how it is used to cross-
validation for model selection and
parameter tuning, kernel methods
Week 15 Bonus lectures (no homework Reinforcement learning
04/17-04/23 required)

Week 16 Final Week Summary of class Project Report Due –

04/24-04/25 04/30.