If you are already involved in a research project, then the topic of your project should be identical to the research you are already doing (e.g., for your PhD research topic). It is also understood that your topic may change over the course of the semester – this is fine. You should write your reports based on the current topic you are working on and do not need to go back to modify prior reports unless noted.

Project Overview

Describe your research project

  • What is the problem you will look at?
  • How do you plan to solve it?
  • Why is the problem important and why will your solution be better than other options?

Total length: ~3 paragraphs, at most 1 page

  • Bring to class or email in PDF format
  • Suggestion: use latex (possibly www.overleaf.com)

Due: September 6th in class

Literature Survey

Doing a literature survey ensures you understand the state of the art work in your problem area and helps identify common issues and opportunities that can motivate your work.

Carefully read N papers related to your project

  • 1 < N < 6 - your choice based on where you are in your project

For each paper:

  • Describe the paper - 1 paragraph
  • List strengths and weaknesses - bullet points with 1-2 sentences each
  • Describe how this work relates to your project/problem

PhD students: Write 1 paragraph on how you will organize papers and track your bibliography. One option is Zotero, but you should pick something you will seriously commit to using so check with others in your research group to see if there is already a normal approach.

Submission: Upload a PDF here. You do not need to bring a copy to class.

Due: Tuesday Oct 5th 11:59PM

Experimental Design

Having a clear experimental design gives structure to your investigation. Having a set of hypotheses to evaluate ensures your solution is robust and by predicting preliminary results you can more easily “sanity check” your experiments to be sure they are behaving correctly.

This report will be formulated as a set of slides giving details of experiments you plan to run. You must use this slide template as a base. From the File menu choose “Make a Copy” and then fill in the slides, or export them to power point. After modifying them you will submit a PDF.

Plan N experiments that will help you understand the problem you are looking at, or your solution.

  • 1 < N < 5 - your choice based on where you are in your project

For each experiment (see Speaker Notes in template for more details):

  • Slide 1 contains: the goal of your experiment, a brief description of how you will run it, and a list of approaches/variations you will use as a comparison.
  • Slide 2 contains: A hypothesis describing what behavior you expect to see, a list of metrics that you will analyze to understand the results, and 1 or more graph sketches that illustrate how you expect those metrics to behave.

The experimental plans in this document do not limit what you evaluate over the coming weeks–you are free to modify your design based on what you discover.

Submission: Upload a PDF here. You do not need to bring a copy to class.

Due: Tuesday October 19th at 11:59PM

Your Approach, Visualized

Every good paper needs at least one key figure that provides a clear and memorable explanation of the work. For this assignment you must create a diagram or graph that will act as a center piece for your project. This could be a system overview diagram or a visual representation of an algorithm. Alternatively, you can make a set of graphs that show the key results of your work, however, you should avoid this unless you are presenting that data in an interesting or unique way – I do not simply want to see some bar or line graphs.

I recommend you make your diagram with diagrams.net, but if your lab consistently uses a different tool, that is also fine. You could also use a tool such as Powerpoint, but in that case you should take care to make it not look like a typical powerpoint diagram. You should follow the advice from Week 10 on how to make attractive, easy to read visuals.

Submission: Upload a PDF here.

Due: Tuesday November 23rd before class.

Project Poster

During our final class we will have a mini poster session. You should create a poster that gives an overview of what you have done this semester. Your poster should:

  • Be easy to read and attractive looking
  • Motivate the problem you are working on
  • Outline your proposed solution
  • Present experimental results or plans

You should create your poster as a single 24x36 inch (landscape) slide in a tool such as PowerPoint/Keynote/Google Slides. The best looking poster will be turned into a template to share with the class.

During our final class, you will give a 3-5 minute research pitch:

  • Overview the problem - why is it important? what is hard?
  • High level description of solution / approach
  • Preliminary results or experimental plans (can be fake)
  • Conclusion that emphasizes key points
  • 3-5 min Q&A (must be at least 3 questions per poster)
  • Can use poster slide directly, or break it into smaller zoomed in slides
    • Don’t make a separate set of slides, pretend that you are standing next to a physical poster and directing our attention to it

Submission: Upload a PDF here.

Due: Tuesday December 7th in class.

Final Report

Combination of all material so far:

  1. Project Overview (~1 page)
  2. Literature Survey (~1-3 pages)
    • Modify based on my feedback if necessary
  3. Proposed Approach (~1-2 pages)
    • New text to explain your visual (~1 page)
  4. Experimental Design/Results (1-2 pages)
    • Describe experiments you either have run or plan to run.
    • Can include “fake” results for what you expect to see (but clearly label them this way)
  5. Conclusion ( 0.5 pages)
    • Emphasize a few key points about your results or your design

Prepare this in a format fitting for your end goal (Latex conference template, Python Notebook, etc).

Note: If you have prepared a conference or journal paper on your research during the last month, then you may submit that in place of this report.