As a recipient of the Foundations of Neuro-Symbolic Multi-Agent Systems studentship, you will receive: 

Neuro-Symbolic systems contain both learning and formal reasoning components. Such a combination of capabilities usually makes them more reliable, accurate and transparent than learning-based systems as well as more adaptable and scalable than symbolic-based systems. Most neuro-symbolic systems are built as single agents designed to solve rather specific tasks, and they tend to perform quite well in those restricted and well-defined environments. For instance, think of an agent that can play better than most humans games such as Chess or Go, but would usually perform very poorly with any other game (or reasoning task).  

In this project, we want to abstract away from the ad-hoc architectures currently existing of neuro-symbolic systems and, instead, develop and further investigate a core architecture, a formal model, for neuro-symbolic systems which can be used to understand the most fundamental mathematical and computational properties of these systems. In this project, we will follow a logic-based and game-theoretic approach that will bring together symbolic reasoning (formal) and learning/probabilistic (data-driven) capabilities, in a natural way. This approach will also allow us to investigate multi-agent extensions of simpler single-agent models, leading to the study of logics for multi-agent systems and multi-agent decision making techniques under uncertainty, as formally done within game-theoretic contexts.  

This project is theoretical in nature but will also validate the designed new models in practice. As such, a strong mathematical background and some programming skills will be required. In particular, the successful candidated will be working in a wide variate of topics in mathematics and computer science, including logic, computational game theory, machine learning, probabilistic reasoning, discrete mathematics, and automated formal verification.  

There will be two related, yet different PhD positions within this overarching project idea: one PhD position is for a candidate mostly interested in logic and formal verification, while the other PhD position is for a candidate mostly interested in mathematical modelling, game theory, and formal techniques for multi-agent decision making. In both cases, the PhD projects fall within the areas of Theoretical Computer Science, AI, and Multi-Agent Systems.  

This scholarship is available to UK and international applicants. 

Applicants are typically expected to hold an upper second-class (2:1) undergraduate honours degree (or equivalent qualification) in a related field. However, those who do not meet this requirement may still be considered if they hold a relevant postgraduate qualification or can demonstrate sufficient professional experience to compensate. 

The University of Sussex believes that the diversity of its staff and student community is fundamental to creative thinking, pedagogic innovation, intellectual challenge, and theinterdisciplinary approach to research and learning. We celebrate and promote diversity,equality and inclusion amongst our staff and students. As such, we welcome applicants from all backgrounds.

Apply online for a full time PhD in Informatics (SEP2025) using our step-by-step guide.