Projects
Click here to register your team with a tentative title, and submit your final report on Brightspace
The goal of doing a project in this course is to explore data privacy by yourself and find something that interests you and make you feel really excited to investigate deeper. You may form teams of one to three people.
I don’t want to place too many constraints on what you do. However, broadly, there are two types of projects: application project and survey projects.
-
In an application project, the idea is to take one or more algorithms from the literature and evaluate them on a new or exsiting dataset or application. Ideally, you might even improve on those algorithms, or adapt them to a slightly different problem.
-
In a survey project, the idea is to summarize ten or more papers all focusing on a specific area, and present one or two of the main technical ideas in depth. Ideally, a theory project will also discover new results that go beyond what is in the literature, or survey papers that were not previously jointly discussed. In the research literature, such publications are often called Systematization of Knowledge (SoK) papers. SoK papers evaluate, systematize, and contextualize existing knowledge. They provide an important new viewpoint on an established, major research area. The heart of the SoK paper is analysis: analyzing the existing literature and providing insights that could not be obtained by simply reading each of the individual papers. SoK papers analyze the current research landscape: identify areas that have enjoyed much research attention, point out open areas with unsolved challenges, and present a prioritization that can guide researchers to make progress on solving important challenges. You can find a list of good SoK paper here.
For both types of projects, the expectations for the scope of the project scale linearly with the number (N) of team members:
-
For application projects, the number of algorithms implemented is N+1.
-
For survey projects, the number of papers to be surveyed is 7*N+1.
Students in a group will be given the same grade. There is no preference from my side on the research format, topics, or whether you work individually or in a group.
A good project consists of a good proposal, a comprehensive survey of existing work within the topic, and a complete execution. At any stage, you are very welcome to send your plan or draft to me for feedback.
You may also choose to do a research project. However, please keep in mind that those are usually more open-ended. If you aim to do a research project and you put a lot of effort into, say, organizing and cleaning existing methods that no one ever did before (which is very important), but you only produce a minor new result, that is probably unfortunate in real research. But it is fine for this class, as long as you clearly document what you have done.
Some ideas for research projects (you are very welcome to come up with your own):
- Explore effective/efficient database reconstruction attacks
- Evaluate/benchmark existing methods for differentially private machine learning
- Privacy attacks and defense on machine learning algorithms
- Building a differentially private system (e.g., an SQL engine or a machine learning training system)
- Interplay between differential privacy and fairness or poison attacks