Syllabus: CS 5043: Advanced Machine Learning (Spring 2018)

Topics

• Decision trees: ensemble methods, random forests, and boosting
• Regression and combating overfitting: ridge regression, Tikhonov regression, lasso, elastic nets, support vector regression
• Nonlinear dimensionality reduction: Kernel PCA, local linear embedding, ISOmap, multidimensional scaling
• Semi-supervised learning
• Deep learning: regularization, convolutional neural networks, recurrent neural networks, autoencoders
• Evaluation in ML: metrics, cross-validation, statistics, addressing the multiple comparisons problem

General Information

• Meeting time: Tu/Th 1:30-2:45

• Location: Carson 121

• Prerequisites:
  • Linear Algebra (Math 3333), AND
  • Statistics (Math 4743 or Math 4753 or ENGR 3293 or ISE 3293), AND
  • Artificial Intelligence (CS 4013/5013) OR Data Mining (CS 5593) OR Intelligent Data Analytics (ISE/DSA 5103) OR Machine Learning (CS 4033/5033).
  OR permission of the instructor.

• Reading Materials:
  • Aurélien Géron (2017) Hands-On Machine Learning with Scikit-Learn and TensorFlow (Concepts, Tools, and Techniques to Build Intelligent Systems), ISBN-13: 978-1491962299 O'Reilly Media,
  • Various papers and other network resources.

• Other key materials:
  • Class schedule: http://www.cs.ou.edu/~fagg/classes/mlfds/schedule.html
  • Scikit-Learn: http://scikit-learn.org/stable/
  • Tensorflow: https://www.tensorflow.org/
  • XG-Boost: https://xgboost.readthedocs.io/en/latest/

• Course web page: http://www.cs.ou.edu/~fagg/classes/mlfds

• We will also be making heavy use of Canvas

• Instructor: Dr. Andrew H. Fagg
  • Office: DEH 243
  • Phone: 325-8606
  • Homepage: http://www.cs.ou.edu/~fagg
  • Email: andrewhfagg--gmail.com
  • Office hours: see the office hours web page http://www.cs.ou.edu/~fagg/office.html

Course Policies

• Attendance: This is a very discussion-oriented course. While keeping up with the readings is an important step to take, it is not a substitute for attending class.

• Class Web Page: Most of the material that you will need can be found on the class web page located at: http://www.cs.ou.edu/~fagg/classes/mlfds

• Canvas: This class will also use Canvas, located at: http://canvas.ou.edu

  Login with your 4+4 (typically the first four letters of your last name followed by the last four digits of your student number), using your standard OU password. If you have difficulty logging in, call 325-HELP. This software provides a number of useful features, including a list of assignments and announcements, an electronic mailing list, newsgroups, and a grade book.

  I may update the main web site and the Canvas page several times a week. When I update the site in any significant way, I will post an announcement on Canvas telling you what has been added and where it is located. You are responsible for things posted on the site within 48 hours of the post.

• Class Communication:
  • The class period will be a mixture of lecture and collaborative project work. Your active participation in both will result in a more salient experience.

  • Outside of class, the discussion group on Canvas should be the primary method of communication. This allows everyone in the class to benefit from the answers to your questions, and provides students with more timely answers.

  • Matters of personal interest should be directed to email instead of to the newsgroup, e.g. informing me of an extended illness.

  • Announcements will be posted to the Canvas announcement board.

  • It is your responsibility to have Canvas configured so that you receive these messages in a timely fashion. Note that Canvas can be configured so that it will forward messages, discussion posts and announcements directly to your email address.

• Final Project: Your final project will occupy the last few weeks of the semester on a topic of your choosing. You will be responsible for an oral presentation for the proposal and a checkpoint, and for a poster presentation on the last day of the semester.

• Proper Academic Conduct:
  • Homework assignments must be your own work.
  • The project will be performed in small groups. Each group member is expected to contribute equally to the final project.
  • In all cases: while the net may be used as a reference, downloading code from the net is prohibited. Note that programs will be checked by software designed to detect improper copying. This software is extremely effective and has withstood repeated reviews by the campus judicial processes.

• Incompletes: The grade of "I" is intended for the rare circumstance when a student who has been successful in a class has an unexpected event occur shortly before the end of the class. I will not consider giving a student a grade of "I" unless the following three conditions have been met.
  • It is within two weeks of the end of the semester.
  • The student has a grade of C or better in the class.
  • The reason that the student cannot complete the class is properly documented and compelling.

• Accommodation of Disabilities: The University of Oklahoma is committed to providing reasonable accommodation for all students with disabilities. Students with disabilities who require accommodations in this course are requested to speak with the professor as early in the semester as possible. Students with disabilities must be registered with the Office of Disability Services prior to receiving accommodations in this course. The Office of Disability Services is located in Goddard Health Center, Suite 166, phone 405/325-3852 or TDD only 405/325-4173.

• Classroom Conduct: Because cell phones and laptops can distract substantially from the classroom experience, students are asked not to use either during class, except in cases in which the laptop is required as part of a classroom exercise.

  Disruptions of class will also not be permitted. Examples of disruptive behavior include:

  • Allowing a cell phone or pager to repeatedly beep audibly.
  • Playing music or computer games during class in such a way that they are visible or audible to other class members.
  • Exhibiting erratic or irrational behavior.
  • Behavior that distracts the class from the subject matter or discussion.
  • Making physical or verbal threats to a faculty member, teaching assistant, or class member.
  • Refusal to comply with faculty direction.
  In the case of disruptive behavior, I may ask that you leave the classroom and may charge you with a violation of the Student Code of Responsibilities and Conduct.

Grades

• In-class participation: 10%
• Homework assignments (6): 40%
• Project work: 50%

General Grade Issues

• Grade questions: If you have a question about grading (including assessment of points), you may address these during office hours or email. Note that if you are asking me to reconsider a grade, then I will likely re-examine the entire project. You have one week from the point that you receive feedback to address grading questions.

• Canvas Grade Summary: Canvas has a grade book that is used to store the raw data that is used to calculate your course grade. It is the responsibility of each student in this class to check their grades on Canvas after each assignment is graded. If an error is found, please bring it to my attention.

Course Evaluations

Adjustments for Pregnancy/Childbirth Related Issues

Title IX Resources

Copyright notice: Many of the materials created for this course are the intellectual property of Andrew H. Fagg. This includes, but is not limited to, the syllabus, lectures and course notes. Except to the extent not protected by copyright law, any sale of such materials requires the permission of the instructor.