Are you interested in developing network science methods and applying them to real-world data? Apply for a PhD position at the Department of Methodology and Statistics of Utrecht University! You will be working on questions such as: how can networks help us understand and predict social systems? How to predict unobserved connections between individuals? What information about individuals is encoded in the network structure?
Social, biological, and economical systems can be represented using networks, where nodes (e.g., people, genes, companies) are connected through relationships (e.g., friendships, regulations, financial transfers). This representation allows us to discover patterns that are unobservable when nodes are studied independently. In the real world, networks are rarely complete, they have missing information on nodes, edges, or metadata. In this PhD project, you will develop methods to understand how much information about node attributes is encoded in the topology of the networks, and how much information about the network topology is encoded in the node attributes.
The corresponding goals of this PhD project are:
- to consolidate the existing knowledge about the relationship between network topology vs node attributes;
- to develop algorithms that can recover attributes using topology, and topology using attributes;
- to understand (and improve) the fairness of these algorithms (understanding and preventing bias), and;
- to apply these methods to real-world phenomena.
The PhD student will develop their own project ideas in coordination with the supervisors and potential partner organizations such as the Anti Money Laundering Centre or Statistics Netherlands. Some examples are:
- Nodes in networks generally have attributes (e.g., type of gene, firm sector). The information of class attributes is partly encoded in the ego-network of the node, and several methods have been proposed to predict the class label in the network based on information from a subset of the nodes. A potential project can explore the extent to which class labels can be inferred solely from the network structure, or by the combination of the network structure and node attributes. A potential application of this project is economic crime detection, where the label of a node (criminal/not criminal) is inferred from the neighborhood of the node.
- The increase in data collection has allowed government and private companies to gather large databases on individuals and their affiliations. A potential project can explore how information on nodes' attributes is embedded in their affiliations, and evaluate the privacy and fairness implications of such algorithms.
- Recent advances in technology and databases have allowed us to track node features (e.g. gene expression, or firm financial accounts) over time. The time series can then be used to reconstruct the (hidden) network topology, allowing us to discover new edges between the nodes representing regulatory interactions. However, the efficiency of the algorithms can be impaired when the time series are correlated due to missing information. A potential solution is to integrate community detection into network reconstruction to account for unobserved factors creating the correlation.
Your work will also include 10-20% teaching tasks. You will be well guided and supported by your three daily supervisors, Dr. Javier Garcia-Bernardo and Dr. Mahdi Shafiee Kamalabad external linkfrom Methodology & Statistics, and Dr. Peter Gerbrands external linkfrom Economics, as well as one senior supervisor (Prof. Dr. Daniel Oberski external linkfrom M&S).
The PhD position is available for 4-4.5 years (depending on 10 or 20% teaching tasks);, with a starting date of September 2022 (negotiable). The PhD student will be appointed to the Department of Methodology and Statistics external linkat Utrecht University, the Netherlands.
- conducting the research (e.g., literature review, developing network algorithms, analyzing data);
- writing international scientific publications and a dissertation that combines the theoretical and statistical aspects of the project;
- giving presentations at (inter)national scientific conferences;
- active participation in the research team of the UU Department of Methodology and Statistics (M&S) and the Department of Interdisciplinary Social Sciences (ISW);
- knowledge utilization: collaborating and sharing findings with practitioners (e.g., professional publications and presentations for partner organizations);
- following courses/training (e.g., statistical and network science courses).