PhD Position Scalable Graph Learning

Graph machine learning (Graph ML) is an emerging field of artificial intelligence (AI) motivated by the ubiquity of graph-structured data in real-world applications. Graph neural networks (GNNs) serve as a key technology in this area and are successfully used in recommendation systems, financial crime analysis, cybersecurity, and social and biological network analysis. However, conventional GNN architectures are limited in the type and complexity of graph patterns they can detect. For example, cycles, cliques, and frequent motifs can serve as highly discriminative signatures when analysing financial, biological, and social networks. However, cost-efficient and scalable discovery of such patterns in large graphs using neural-network-based approaches is challenging. Although these patterns can be detected using purely combinatorial approaches, such approaches lack statistical learning and adaptation capabilities and have high computational complexity.
A key objective of this project is to identify the trade-offs between combinatorial and neural-network-based approaches in the Graph ML space. We aim to combine and unify combinatorial algorithms and GNNs based on linear-algebra-based primitives, and achieve more efficient and scalable solutions. Our goal is to answer questions like the following: 

• What capabilities are needed in GNNs to solve combinatorial graph problems?
• How can combinatorial graph algorithms and GNNs complement each other?
• Can such solutions be offloaded to AI accelerators such as GPUs and TPUs?
• Can such solutions be executed on large graph datasets in a distributed fashion?
• Can such solutions serve as effective countermeasures against financial crime?

The project will cover some real-world applications of Graph ML in the financial space, including analysis of financial transaction networks to detect criminal activities such as money laundering and fraud. In this area, we will explore new ways of combining GNNs and Large Language Models (LLMs) to construct machine learning solutions that can operate on graph-structured data with rich sets of node and edge features. 

Requirements:

We are looking for a candidate who satisfies the following requirements:

• an MSc degree with excellent results in Computer Science, Mathematics, Electrical Engineering or a related discipline with an MSc thesis conducted preferably in the field of graph theory, machine learning, deep learning, or parallel & distributed computing,
• a solid background in algorithms and complexity analysis,
• strong programming skills and practical experience with Python and C/C++,
• hands-on experience with Deep Neural Networks using PyTorch or TensorFlow,
• good speaking and writing skills in English with a minimum TOEFL-score of 100 or a minimum IELTS score of 7.0 per sub-skill (writing, reading, listening, speaking), which applies to all candidates wanting to pursue a PhD or PDEng programme at TU Delft.

Doing a PhD at TU Delft requires English proficiency at a certain level to ensure that the candidate is able to communicate and interact well, participate in English-taught Doctoral Education courses, and write scientific articles and a final thesis. For more details please check the Graduate Schools Admission Requirements.

Salary Benefits:

Doctoral candidates will be offered a 4-year period of employment in principle, but in the form of 2 employment contracts. An initial 1,5 year contract with an official go/no go progress assessment within 15 months. Followed by an additional contract for the remaining 2,5 years assuming everything goes well and performance requirements are met.
Salary and benefits are in accordance with the Collective Labour Agreement for Dutch Universities, increasing from € 2872 per month in the first year to € 3670 in the fourth year. As a PhD candidate you will be enrolled in the TU Delft Graduate School. The TU Delft Graduate School provides an inspiring research environment with an excellent team of supervisors, academic staff and a mentor. The Doctoral Education Programme is aimed at developing your transferable, discipline-related and research skills.
The TU Delft offers a customisable compensation package, discounts on health insurance, and a monthly work costs contribution. Flexible work schedules can be arranged.
For international applicants, TU Delft has the Coming to Delft Service. This service provides information for new international employees to help you prepare the relocation and to settle in the Netherlands. The Coming to Delft Service offers a Dual Career Programme for partners and they organise events to expand your (social) network.

36 - 40 hours per week

Mekelweg 2