CSCI 6220 Research Project, Fall 2026

The research project gives graduate students experience reading and engaging with the randomized algorithms research literature. Projects are completed in pairs.

Scope. Your project must relate to randomized algorithms. Acceptable directions include: extending an algorithm from the course (e.g., with empirical evaluation, tighter analysis, or a novel variant); surveying and critically comparing several related randomized techniques from the recent literature; or beginning a novel research direction. Straightforward application of an existing randomized method to a dataset, without algorithmic or analytical depth, is not acceptable.

Format. The project consists of four graded components: topic selection, a progress report, final deliverables, and an in-class presentation.

Grading and Deadlines

Task Due % of project grade Details
Topic selection Lecture 6 10% One-paragraph proposal submitted via Submitty: what topic, why it relates to the course, initial references.
Progress report Lecture 15 30% A partial report (2–3 pages) including: a clear problem statement, literature review, planned approach, and any preliminary results. Should demonstrate substantive progress.
Final deliverables Lecture 25 25% Final written report (4–6 pages, typeset using ICLR workshop format) and a public GitHub repository with well-documented code to reproduce any empirical results.
Presentation Lecture 26 35% Prerecorded 20-minute presentation played in class. Every group member must speak. Upload your recording and slides to Submitty before your scheduled presentation date.

Topic Selection

Submit a one-paragraph proposal via Submitty describing: the randomized algorithm or technique you will study, why it relates to the course material, and 2–3 initial references. The instructor will respond with approval or guidance for refinement. Your topic must be approved before you proceed to the progress report.

If you are unsure where to start: browse recent STOC, FOCS, SODA, ICALP, or RANDOM proceedings for papers on randomized algorithms that interest you. Alternatively, consider exploring a topic covered briefly in lecture (e.g., cuckoo hashing variants, extensions of the probabilistic method, concentration inequalities for dependent variables) in greater depth.

Progress Report

The progress report is a partial draft (2–3 pages) of your final report. It should include:

Expect the progress report to represent roughly half of the total project effort. If you encounter difficulties, meet with the instructor well before the deadline.

Final Deliverables

Report. A 4–6 page paper typeset in ICLR workshop format (see ICLR template). The report should present your topic clearly, with proper mathematical typesetting, a coherent narrative arc (problem → approach → results → discussion), and complete references.

Code. A public GitHub repository containing well-documented, cross-platform code to reproduce any empirical results. Include a README with instructions for setup, data acquisition, and reproduction. If your project is purely theoretical, the repository should instead contain a self-contained PDF of your report and any supplementary computations (e.g., a Mathematica or Python notebook verifying algebraic steps).

Presentation

Presentations are 20 minutes, played in class at the end of the currently available lecture sequence. Address the following to receive full credit:

Slides must use properly typeset mathematics and legible figures. Submit your recording (uploaded to RPI Box or equivalent) and slide deck to Submitty before your scheduled date.

Resources

For guidance on giving effective talks and reading research papers, see the slide decks from the 2019 CS Grad Skills Seminar (Prof. Anshelevich and Prof. Xia).