Your class project offers an opportunity to apply probabilistic graphical models (PGMs) and modern probabilistic AI techniques to real-world problems. Projects will be completed in teams of 2-4 students and are worth 30% of your final grade, consisting of the following deliverables:

  1. Proposal (5%): 2 pages, excluding references.
  2. Midway Report (5%): 4 pages, excluding references.
  3. Presentation (5%): In-class presentation.
  4. Final Report (15%): 8 pages, excluding references.

All submissions must adhere to the ICML style template.

Team Formation

Teams of 2-4 students are strongly encouraged. Once your team is formed, email the instructor with the names of all members. If assistance is needed, contact the instructor to help find project partners.

Deliverables and Deadlines

Proposal

Due: Friday, March 7, 2025, at 11:59 PM via Canvas.

  • Content:
    • Project title and team member list.
    • Problem statement and motivation (½ page).
    • Literature review of at least four relevant papers (~1 page).
    • Description of dataset(s) and planned activities.
  • Grading:
    • 40%: Clear and concise description of the project.
    • 40%: Quality of literature survey.
    • 10%: Feasibility and detail of activity plan.
    • 10%: Writing quality.

Midway Report

Due: Friday, April 11, 2025, at 11:59 PM via Canvas.

  • Purpose: This report ensures you’ve made real progress on your project. By now, you should have:
    • Implemented an initial version of your model or experiments.
    • Run at least some tests—even if they didn’t work.
    • Identified challenges and next steps. This is not just a plan—it must reflect actual work done so far.
  • Format & Length:
    • 4 full pages, excluding references. May slightly spill onto a fifth page, but be concise.
    • Use the ICML template (single-column format).
    • References must be on a separate page.
    • Figures & Tables
      • Use figure for single-column figures, figure* for full-width figures. Figure sizes (including using \width= commands) will significantly change your report length, so size them properly.
      • Ensure figures are numbered, referenced, and captioned properly.
    • Required Sections (length of each sections will vary by project focus)
      • Abstract (~150-200 words)
        • Summary of the problem, approach, progress, key challenges, and next steps.
      • Introduction (½ - 1 page)
        • Problem statement and motivation.
        • Role of probabilistic graphical models (PGMs) or probabilistic AI.
      • Related Work (~½ page)
        • Summary of at least four relevant papers and how they relate to your work.
      • Methods (~1 page)
        • Implemented work so far (not just plans).
        • Model structure, key assumptions, modifications.
        • Challenges encountered and adjustments made.
      • Experiments (~1 page)
        • Dataset(s) and experimental setup.
        • Preliminary results, even if unsuccessful.
      • Conclusion & Updated Plan (~½ page)
        • Summary of progress, challenges, and next steps.
        • Timeline for final report.
  • Grading
    • 20%: Abstract, Introduction & Related Work (Clarity, relevance, citations)
    • 30%: Methods & Implementation (Actual work done, challenges, technical clarity)
    • 30%: Experiments & Preliminary Results (Progress, results, adjustments)
    • 10%: Next Steps & Timeline (Realistic plan, addressing issues)
    • 10%: Writing & Organization (Clarity, structure, proper citations & figures)

Final Report

Due: Monday, May 5, 2025, at 11:59 PM via Canvas.

  • Content:
    • ICML-style paper (8 pages).
    • Sections: Introduction, Background & Related Work, Methods, Experiments, and Conclusions.
  • Grading:
    • 10%: Problem definition and literature review.
    • 30%: Methodology (soundness and originality).
    • 30%: Experiments and results.
    • 10%: Analysis and discussion.
    • 20%: Writing quality.

In-Class Presentation

Date: Tuesday and Thursday, April 29 and May 1, 2025.

  • Format: Each team will deliver a 8-minute presentation in class, summarizing their project. Presentations will include a brief introduction, methods overview, key findings, and future directions.

  • Grading:

    • 30%: Clarity and engagement of presentation.
    • 40%: Depth and quality of content.
    • 30%: Use of visuals and adherence to time limits.

Suggested Project Topics

See slides from Lecture 7 for project ideas.