Midterm presentations (updated with handout weblinks)

Schedule

Presentations will take place on Monday 16 and Wednesday 18 March (right after midterm break).

Choice of topic

Most topics related to mathematical aspects of AI are admissible:

  • The topic should be an extension of concepts in the course; it should not be a syllabus topic (unless understood in a broader and deeper sense).
  • The mathematical content must be significant (challenging but not too advanced considering your standing as undergraduate or graduate student). Theoretical/foundational topics as well as applied topics (as in applied math) are acceptable.

It is highly recommended (but not required) to discuss your choice of topic with the instructor.

Format of presentation

The presentation will be oral.

  • I discourage “blackboard-only” presentations unless you have significant experience (e.g., have given several other similar presentations, or maybe you are a seasoned tutor and good explaining in real time). –>
  • You are encouraged to be creative. Deep learning has copious concepts that visual explanation: include at least one picture or diagram, show a video (but very short —no more than 1 minute long!), or prepare a relevant Colab (with at least an image or two) to include as part of the presentation. A purely blackboard-based presentation is possible, but not particularly encouraged (if you do a blackboard-only presentation, make double-sure that your handout includes figures or diagrams too detailed to fit into the short format).
  • Your presentation should last 8 to 10 minutes.
    • The 10 minutes limit is absolute; you should plan and aim for an 8-minute presentation. There is no credit to gain by exceeding 8 minutes —in fact, it is harder to give a good, short and complete presentation than a longer one (which may affect your credit).
    • Any presentation exceeding lasting over 10 minutes will be subject to a 10% credit deduction for each additional minute (or fraction) exceeding the hard 10-minute limit. (This is necessitated by the need to fit a dozen presentations into 2 class meetings.)
  • At the end of your presentation, there will be a 1–to-2 minute (maximum) interval for questions (Q & A) from the audience related to —or perhaps extending— the material presented. You should prepare accordingly!
    • If your presentation is under 9-minutes long, you will have a full 2 minutes for questions; if it is over 9 minutes long, the excess will decrease your allotted Q & A time.
    • If no other student has a question, there will be an automatic 10% deduction. (In this case, the instructor will ask at least 1 question.) It is in your best interest to give a clear and engaging presentation that will invite and spur student questions!

Accompanying handouts/supplements

Your presentation must be accompanied by a handout (1 to 2 pages in length), meant to accompany and provide references to the material presented. Ideally, the handout will contain some figure(s) to accompany your presentation (particularly if you give a mostly- or purely-blackboard presentation).

Be creative! As an example (not a requirement!), your handout may include a QR code to a file shared in you Google drive (or Dropbox, or OneDrive…) containing the handout itself in electronic form, or a document or webpage where you put materials (e.g., links to videos presented, to web references, to a shared Colab, etc.)

Grading

  • [10%] Choice of topic (interesting, relevant, of suitable sophistication for an upper-level undergrad/grad student).
  • [40%] Technical content (correctness, coherence, completeness).
  • [25%] Delivery (manner of execution of presentation: format, polish, flow, audience engagement)l
  • [25%] Handout (originality, suitability, quality of execution

Bonus: Up to +10% credit. Include a couple (or, at least, a good one) of “hands-on”/exploratory/inviting/challenging questions or activities for the reader. If your activity includes significant coding of your own, up to another +10% may be earned; e.g., create and share a Colab program (in Python/NumPy/Keras…) to illustrate some concept and integrate into a proposed activity.

Again, be creative!