Project

The goal of the project in this course is for you to independently explore data privacy, find a topic that excites you, and dive deeper into it. Through this process, you will learn how to conduct a comprehensive review, manage your time effectively, and write detailed reports.

You are free to use any format for your project deliverables. Whether it is in LaTeX, Google Docs, or any other medium, the key is to convey your findings effectively. There are three types of projects you can choose from:

• Research: A research project aims to introduce new ideas or concepts.
• Evaluation: An evaluation project involves conducting a thorough evaluation and comparison of a particular topic, such as summarizing and evaluating existing algorithms for a specific problem.
• Application: This type of project focuses on applying existing techniques to a new domain or dataset in a meaningful and non-trivial way.

If you choose a research project and invest a lot of effort into organizing and analyzing existing methods—even if your final result is a minor new contribution—that is completely acceptable for this course. Similarly, finding new insights during an evaluation project is highly valued. Please make sure to clearly outline your contributions from both perspectives.

You can work individually or in a group of up to two people, but note that the expected amount of work will double for a group project. Group members will receive the same grade. There is no preference from my side regarding project format, topic, or whether you work alone or in a group.

A successful project will include a well-thought-out proposal (your topic may evolve, but the more planning you do early on, the better), a comprehensive survey of related work, and thorough execution. There are three key milestones for the project:

• Proposal: A detailed project proposal outlining your topic, objectives, and plan of action. This will help define your focus and get early feedback.
• Poster: A visual summary of your work, presented towards the end of the course, to communicate your findings effectively to your peers.
• Final Report: A comprehensive written report that documents your entire project, including your methodology, results, and insights. At any stage, you are encouraged to seek feedback from the instructor or TA to ensure you stay on track.

---

Some ideas (and you are very welcome to come up with your own idea):

• Evaluate/benchmark existing methods for differentially private machine learning
• Incorporate public information during differentially private data release
• Differential privacy into new machine learning algorithms (I am not an expert of machine learning; I only know image CNN, NLP, GNN; if you know more and want to apply differential privacy to some machine learning paradigm that no one did before, that would be great)
• Privacy attacks of new machine learning algorithms
• Generating differentially private code automatically
• 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