Do you want to trace back a computer virus attack or find patient zero of an epidemic? Do you want to detect where the power grid was disrupted? Do you want to find out who spread the rumor in your social network? Or which of your molecules an unknown substance is acting upon?
All these problems can be formulated as one of source detection in complex graphs. But which algorithm works well for your application? This probably depends on the type of network as well as the type of information spread, and whether you need short runtime or high-quality results. Let’s find out together!
In this software project, different teams will work on the goal of implementing and benchmarking methods of source detection. We will have teams that implement existing supervised and unsupervised methods, come up with new algorithms, and create benchmarking datasets and metrics. We aim to make our methods fit for applying on real-world problems and prepare a publication draft.
In this project, in addition to tackling an interesting and complex prediction problem, you will learn how to consider interfaces between teams, train organizing yourselves, coding scientifically and reproducibly, and interacting with scientific literature.
Depending on the audience, we can also work in German.
Our first meeting will be on Tuesday, April 18, 15:15-16:30. We plan to offer the seminar in a hybrid format, i.e., we plan to be in the lecture halls (HPI Campus Griebnitzsee, K.1.02, or at the FU campus) for the weekly meetings and make live dialing-in via zoom available. Please send me an email (katharina.baum@hpi.de) if you want to attend, so I can share the dial-in information and add you to the course’s moodle where all other information is supplied.
Teaching and Learning
The majority of the project work will consist of hands-on project work that includes programming, data preparation and analysis, result interpretation, visualization, and reporting; you are allowed to work in pairs on your project.
The first two meetings will be held in a lecture-like format to recapitulate relevant basics. Teams will then form and choose their preferred subproblem. We will meet weekly with all teams to discuss about progress and common interfaces for data and information transfer between teams, these meetings will have a highly interactive character. Additional project-specific meetings are possible on demand. During the last meetings, you will present your project and its results in a final talk that covers your whole analysis, and you will be asked to hand in a written report as well as your documented code.