Overview
For CS 686 students, the project is
- required.
- individual.
- worth 20% of the final grade.
For CS 486 students, the project is
- optional.
- done in a group of up to 3 students.
- worth 5% bonus on top of the final grade.
Project Scope
In this project, you can choose from one of several report types below.
For all report types, you must first identify a problem that you want to address and survey 8-10 papers in the literature that tackle this problem.
Afterwards, you can choose to
-
Write a survey apper. The goal of a survey paper is to summarize and critically reflect on the prior work in the literature on a particular topic. Purely summarizing several papers is not sufficient. Your paper must answer two big questions: What have we done so far? What should we do in the future?
-
Choose two existing algorithms and evaluate the performance of the two algorithms on problem instances. For this type of report, you may be reproducing existing results. Choose specific instances of the problem and implement at least two algorithms to solve these problems. Afterwards, discuss several questions. Which algorithm performs better? Does this match the results from prior work? Do the chosen problem instances present challenges that are not addressed by the algorithms? How did you resolve these challenges? Can you propose future directions based on your result?
-
Propose a new algorithm and evaluate its performance on problem instances (by comparing with an existing algorithm). Choose specific instances of the problem that cannot be fully addressed by existing algorithm. Propose a new algorithm to solve these problem instances and compare the performance of your new algorithm with at least one existing algorithm.
-
Propose a new algorithm and evaluate its properties theoretically (by comparing with an existing algorithm). The steps are almost the same as the previous report type, except that you would be comparing the performance of your new algorithm and an existing algorithm theoretically (likely using proofs).
Grading
The project is graded out of 100%. To encourage you keep on track, the project consists of three deliverables.
- Proposal. Due at 6 pm on June 7, 2019.
- Milestone Report. Due at 6 pm on Thursday, July 18, 2019.
- Final Report. Due at 6 pm on August 9, 2019.
You must submit all the deliverables to have a non-zero grade for your project.
Proposal
Milestone Report
Final Report
- Due at 6 pm on August 9, 2019 on Learn.
- Worth 75% of your project grade.
- Report can be up to 8 pages excluding references.
- Use this template. final-report.pdf final-report.zip
Resources
Many of the AI conferences and journals post their proceedings online so that they are accessible for everyone (and the few that do not will be accessible if you are on campus or using the library proxy).
Some resources that you might find useful include
- Association for the Advancement of Artificial Intelligence (AAAI)
- International Joint Conference on Artificial Intelligence (IJCAI)
- Annual Conference on Neural Information Processing Systems (NIPS)
- International Conference on Machine Learning (ICML)
- International Conference on Autonomous Agents and Multiagent Systems (AAMAS)
- Association for Computational Linguistics
- ACM Special Interest Group on Artificial Intelligence (SIGAI)
- Journal of Artificial Intelligence Research (JAIR)
- Artificial Intelligence Journal (AIJ)
This list is far from complete and there are many other relevant conferences and journals that you can consult.
Questions
If you have any questions, or are stuck for a project idea then please contact the instructor and the TAs. We are more than happy to chat about research ideas.