Advanced topics in optimization: From simple to complex ML systems
| Email (instructor): anastasios@rice.edu | Web: https://akyrillidis.github.io/comp545/ |
| Email (course): RiceCOMP545@gmail.com | |
| Office hours: By appointment | Class hours: T\TH 15:10 - 16:30 |
| Office: DH 3119 | Classroom: Online |
| Course Syllabus | LaTEX template for scribing |
| Course description | Schedule | Grading policy | Literature |
Course format and structure
There will be a traditionally formatted series of lectures via Zoom. Students are supposed to present papers after the end of “chapters”, presenting recent advances on the topics discussed. During lectures or presentations, participation with questions / comments is encouraged. The papers to be presented will be selected from a pile of papers that are related to the topics of the course and will be provided by the instructor.
The grade is based on the following factors:
- 5% participation and attendance.
- 25% presentations
- 70% final project
The instructor reserves the right to curve the scale dependent on overall class scores at the end of the semester. Any curve will only ever make it easier to obtain a certain letter grade.
Scribing notes (bonus)
Every week, a different student (or group of students) can volunteer (otherwise, will be randomly selected) to take notes and prepare a short –but consistent– note on the material presented each week. A latex template will be shared by the instructor.
Final project logistics
Students can team up (up to 3 members per group). The goal of the project is to engage students to research related topics, even beyond the timeframe of the course. I.e., there can be topics for a project that focus on simpler scenaria (say convex optimization), and topics that consider some harder non-convex questions. While the former could be potentially finished during the timeframe of the course, the latter could continue after the end of the course (this is the meaning of a open-ended project), and the instructor ``bets’’ (and believes) on the self-motivation of the students to continue working on it, after the end of the course.
A project must include:
- The study of at least 3 research papers, on which the project is based on.
- The proposal of at least one “open” question: this includes the theoretical analysis of a specific scenario, or the implementation of a ML/AI system for some task in an interesting scenario, or a survey comparison of several algorithms on an interesting task.
The instructor will provide feedback to the students (by appointments + electronic communication). There will be a “midterm” 5-minute (tentative) pitch from each group (a session will be booked for this purpose - dates are tentative). After the discussion with the instructor, each group should prepare a four-paged description of the project with:
- Abstract and Introduction.
- Description of state of the art (summary and connection of the 3 papers selected, description of strengths and weaknesses and how these have led to the open question).
- Any preliminary results you have, and what is the plan from now on.
The project will culminate in a final project report of at least six pages, not including references, in NIPS/ICML format. At the end of the course, the group will prepare a 10-15 minute presentation, describing the background and the results they obtained. Final report dates will be available towards the end of the semester.