CS 49/149: 21st Century Algorithms, Fall 2018 : Project Page
Goal:
The purpose of the project is to deeply understand some recent paper relevant to the course. A list of possible papers is here .
To see if a paper is still available, see here .
Overview and Deliverables
Project should be done in teams of two (I will allow at most one solo or 3-member project in case parity forces me to).
There are three deliverables in the project:
- a one-page proposal (10%)
- a 5-10 page report, due 24hrs before your presentation (20%)
- a 25 minute in class presentation (20%)
The team should pick a relevant paper and perform an in-depth analysis. There is a list given below of candidate papers, but you are free to choose one yourself (but run it by me).
The objective is to tease out the main algorithmic/conceptual ideas and techniques.
Proposal
To help you get started early, I require a one-page project proposal from each team by September 30th . The proposal should include
- Team Members
- Paper(s) you would be studying
- A paragraph on what made you choose this paper.
- A paragraph on what you think you would do in the project. You can (should) be ambitious here; I want to make sure you have spent at least sometime
thinking about it.
Note: You are allowed to choose a paper on a topic which we have not yet covered. If you are planning to choose a paper not in the list below, please let me know well in advance.
You are also encouraged to meet me during office hours to discuss your selection.
But remember -- 30th is a hard deadline, and it is 20% of the total project points!
The Report
This should then be distilled into the report (which should be material for say a 2 hour lecture).
Here are some possible examples of questions that your report should address:
- What is the main problem that the paper addresses?
- Why is the problem important and interesting? You are encouraged to critique the paper.
- What were the previous approaches (if any) to the problem?
- What is the main result of this paper?
- What was your "a ha!" moment while reading the paper?
- What are the new questions that arose in your mind while reading the paper, and what are your (partial) answers to them?
The above questions form a guideline, but you should pursue whatever really fancies you. For instance, you could implement the algorithm in the paper and see
if the theory is consistent with the experiments (if the paper says we improve the results of so-and-so, is that something seen in real data?)
Or, you could take a different angle at the question asked in the paper, and try finding out answers. The report will be due 24 hours before your presentation.
The Presentation
The presentation is a further distilling of your understanding/findings to give a 25 minute presentation. Note this is the usual time allotted for talks at a conference,
and it will force you to think what was really cool and important. You will be giving talk in the last week of classes.
Evaluation
Your presentation will be evaluated by your peers (which will count towards their participation marks). Your report will be evaluated by me.
Some Possible Papers