This is the syllabus for CSC580: Principles of Machine Learning.
The University of Arizona is moving from in-person instruction to remote instruction as part of the COVID-19 mitigation efforts.
Specifically, this will impact CSC580 in the following ways:
Please take your health as your first priority. If you foresee that COVID-19 will interfere with your ability to attend class or finish homework assignments, please contact me at your earliest convenience to discuss it.
Yes, I can write you a letter, but I will not write a letter for you simply based on your course performance: transcripts are already good indicators of course performance. If you intend me to write a letter because of a course I taught you, you need to talk to me at the beginning of the semester (in this specific case, before the end of January 2019), and you should read this section on my advice page.
Students will learn why machine learning is a fundamentally different way of writing computer programs from traditional programming, and why this is often an attractive way of solving practical problems. Machine learning is all about automatic ways for computers to find patterns in datasets; students will learn both advantages and unique risks that this approach offers. They will learn the fundamental computational methods, algorithms, and perspective which underlie current machine learning methods, and how to derive and implement many of them.
Students will learn the fundamentals of unsupervised and supervised machine learning methods, the computational and quality tradeoffs between different methods, and how to adapt existing methods to fit their own research needs.
Videoconference lectures, individual assignments, written exams, projects, group discussions.
A successful student will be able to implement and explain the limitations of many of the central methods and techniques in machine learning:
For a more granular description of the learning objectives, see the week-by-week schedule and the description of the assignments below.
Machine Learning is a big field, and there is no way we can cover all of it in one course. With that said, this course covers a large amount of material, and the assignments are a central part of the course. Students are expected to dedicate a significant amount of time on the course outside of the classroom, especially if they have background deficiencies to make up.
The UA’s policy concerning Class Attendance, Participation, and Administrative Drops is available here. The [UA policy regarding absences for any sincerely held religious belief] (http://policy.arizona.edu/human-resources/religious-accommodation-policy), observance or practice will be accommodated where reasonable. Absences pre approved by the UA Dean of Students (or dean’s designee) will be honored. See the dean of students’s website for details.
If you register late for this class, contact me as soon as you do. You will be expected to submit all missed assignments within a week of your registration. It is solely your responsibility to catch up to the class content.
We will use Piazza for communications and discussion. Make sure your Piazza account is up to date - class announcements are sent through the website.
The required textbook is Hal Daumé’s Course in Machine Learning, fully and freely available online.
As mentioned above, you will be assessed based on your performance on programming assignments, one midterm exam, one final exam, and one project.
The instructing staff will grade your assignments, project, midterms, and final exam on a scale from 0 to 100, with the following weights:
Your final grade in the course of be the best of a per-class grading curve and overall performance:
By your last day to withdraw, you will know more than 40% of your grade by weight.
Grades for assignments, midterm and final exam will be posted on D2L as soon as we have them. The grading for each assignment will be provided one week after the assignment is due.
There will be a total of 11 programming assignments paced at around one assignment per week, skipping weeks for the midterm, break, and final. Each assignment will be due at least one week after it is posted. The final project report is due at the last day of class.
2020-03-17: The midterm exam will now be a take-home exams given on Mar 23rd 9:30AM. It will be an exam expected to take 75 minutes to complete, and students will be asked to take the exam at any contiguous period between 9:30AM and 11:59PM on that day. The final exam will have the same format, except it will be a comprehensive exam that is expected to take 120 minutes to complete, with a date currently TBD.
The (approximately weekly) assignments are described in the list of scheduled topics below.
The written assignments will serve to make sure students have a firm grasp of the theoretical foundations on which the course material is built, and will explore important variations of topics which were discussed in class.
The programming assignments, on the other hand, serve two purposes. First, they will provide the students an opportunity for hands-on experience with the kind of practical data science knowledge that is deeply relevant in applications:
Second, the programming assignments will provide students an opportunity to understand how popular ML libraries are actually implemented, and why.
|A1||Decision Trees (programming assignment)|
|Students implement a procedure for creating decision trees and decision forests|
|A2||Naive nearest neighbors, k-means (programming assignment)|
|Students implement nearest-neighbor classification and the k-means clustering algorithm|
|A3||Project Proposal (written assignment)|
|Students submit a proposal for their final project, identifying the paper they propose to implement, the datasets they will use for the experiments, and the strategy they will use to evaluate their results.|
|A4||Perceptron and Feature Selection (programming assignment)|
|Students implement the classic linear perceptron algorithm and use methods such as cross-validation to pick one among a number of candidate models based on which features are selected.|
|A5||Reductions (programming + written assignment)|
|Students derive reductions from single-class classifiers to create multiclass classifiers and implement the reductions using the classification software they developed from previous assignments.|
|A6||Linear and Kernel Methods (programming + written assignments)|
|Students derive the optimization procedures for linear least squares and the different formulations of the convex problem that underlies support vector machines, and implement a procedure to optimize both functionals.|
|A7||Naive Bayes (programming assignment)|
|Students implement a basic spam filtering algorithm based on Naive Bayes and evaluate it on existing email corpora.|
|A8||Bias and Fairness (programming assignments)|
|Students observe how their own software can generate structurally-biased predictions and implement procedures to assess and mitigate these issues based on the literature|
|A9||Neural Networks and Back-Propagation (programming assignment)|
|Students implement reverse-mode automatic differentiation, use it to automatically synthesize back-propagation algorithms for a number of neural network configurations, and use these algorithms to train neural networks to classify images of digits and clothing items|
|A10||Ensembling and Efficiency (programming assignment)|
|Students implement stochastic gradient descent for the loss function in A9, and boosting for the models in A1|
|A11||Computational Learning Theory (written assignment)|
If you wish to dispute your grade for an assignment, midterm or project, you have two weeks after the grade has been turned in. In addition, even if only you dispute one portion of the grading for that unit, I reserve the right to revisit the entire unit (assignment, midterm, or project).
The scheduled assignments will present an opportunity for the students to develop important skills associated to machine learning, specifically about exploratory data analysis and data cleaning. Some of the assignments will require students to convert input data from a raw format to something more appropriate for the specific task. Students are expected to be familiar with UNIX command-line tools, and will be encouraged to develop their homework assignments so they can be executed in a shell.
The final project for the course will involve the implementation of a recently-published machine learning technique of the student’s choice, subject to the approval of the instructor. The instructor will provide a list of suggested techniques for students who prefer not to make the choice from scratch. The instructor expects these techniques to be drawn from top research venues in the field, such as NIPS, ICML, ECCV, ICCV, ICLR, etc.
The Department of Computer Science is committed to providing and maintaining a supportive educational environment for all. We strive to be welcoming and inclusive, respect privacy and confidentiality, behave respectfully and courteously, and practice intellectual honesty. Disruptive behaviors (such as physical or emotional harassment, dismissive attitudes, and abuse of department resources) will not be tolerated. The complete Code of Conduct is available on our department web site. We expect that you will adhere to this code, as well as the UA Student Code of Conduct, while you are a member of this class.
To foster a positive learning environment, students and instructors have a shared responsibility. We want a safe, welcoming, and inclusive environment where all of us feel comfortable with each other and where we can challenge ourselves to succeed. To that end, our focus is on the tasks at hand and not on extraneous activities (e.g., texting, chatting, reading a newspaper, making phone calls, web surfing, etc.).
Students are asked to refrain from disruptive conversations with people sitting around them during lecture. Students observed engaging in disruptive activity will be asked to cease this behavior. Those who continue to disrupt the class will be asked to leave lecture or discussion and may be reported to the Dean of Students.
Some learning styles are best served by using personal electronics, such as laptops and iPads. These devices can be distracting to other learners. Therefore, students who prefer to use electronic devices for note-taking during lecture should use one side of the classroom.
The UA Threatening Behavior by Students Policy prohibits threats of physical harm to any member of the University community, including to oneself. See http://policy.arizona.edu/education-and-student-affairs/threatening-behavior-students.
This course will, at times, contain material of a mature nature, which may include references to historical violence as collected and depicted in datasets and historical events. The instructor will provide advance notice when such materials will be used. Students are not automatically excused from interacting with such materials, but they are encouraged to speak with the instructor to voice concerns and to provide feedback.
At the University of Arizona we strive to make learning experiences as accessible as possible. If you anticipate or experience physical or academic barriers based on disability or pregnancy, you are welcome to let me know so that we can discuss options. You are also encouraged to contact Disability Resources (520-621-3268) to explore reasonable accommodation.
If our class meets at a campus location: Please be aware that the accessible table and chairs in this room should remain available for students who find that standard classroom seating is not usable.
Students are encouraged to share intellectual views and discuss freely the principles and applications of course materials. However, graded work/exercises must be the product of independent effort unless otherwise instructed. Students are expected to adhere to the UA Code of Academic Integrity as described in the UA General Catalog. See http://deanofstudents.arizona.edu/academic-integrity/students/academic-integrity.
The University Libraries have some excellent tips for avoiding plagiarism, available here.
The University is committed to creating and maintaining an environment free of discrimination; see the policy document.
Our classroom is a place where everyone is encouraged to express well-formed opinions and their reasons for those opinions. We also want to create a tolerant and open environment where such opinions can be expressed without resorting to bullying or discrimination of others.
UA Academic policies and procedures are available here.
Student Assistance and Advocacy information is available here.
Campus Health information may be found here.
OASIS Sexual Assault and Trauma Services: http://oasis.health.arizona.edu/hpps_oasis_program.htm.
Please see http://www.registrar.arizona.edu/personal-information/family-educational-rights-and-privacy-act-1974-ferpa?topic=ferpa for information on confidentiality of student records. This has concrete consequences for you if you give my name as a reference! In other words, if you intend to give my name as a reference, please contact me ahead of time so we can discuss. To give a concrete example: before you give me your permission, I’m not allowed to discuss your in-class performance with anyone else! This means that if your potential employer is not aware of the issue, they might take my “I cannot answer because of FERPA” answer to mean “I’m not comfortable sharing the student’s performance”, when I actually don’t mean that.
Information contained in the course syllabus, other than the grade and absence policy, may be subject to change with advance notice, as deemed appropriate by the instructor.