Primary supervisor

Helen Purchase

Co-supervisors

Communication networks show interaction between people over time, and are key to the identification of criminal networks and criminal activity. This project will investigate how future events might be able to be predicted, based on dynamic graphs representing prior interpersonal communications. The project will consider (a) how Graph Neural Networks can best be used for this machine learning task; (b) how visualisation techniques can best depict both known-past and predicted-future events.

Aim/outline

The expected outcomes of the project are:

- a literature review and synthesis that details how Graph Neural Networks can be used for this event prediction task

- a machine learning system that takes as input a dynamic communication network and produces a set of predictions (using Graph Neural Networks)

- a visualisation system that visualises both the input and the output

- preliminary evaluation of the tool for both accuracy and usability

URLs/references

A.E.Sizemore, D.S.Bassett. Dynamic graph metrics: Tutorial, toolbox, and tale. NeuroImage 180 Part B 417-427 (2018) 

C.Wilson, J.Dalins, G.Rolan. Effective, Explainable and Ethical: AI for Law Enforcement and Community Safety. IEEE / ITU Int. Conf. on Artificial Intelligence for Good 186-191 (2020)

S.M.Kazemi et al. Representation Learning for Dynamic Graphs. Journal of Machine Learning Research 21, 1-73 (2020)

D. Archambault, H. Purchase, B. Pinaud. Animation, small multiples, and the effect of mental map preservation in dynamic graphs. IEEE TVCG, 17:4 (2010)

J.You, T.Du & J.Leskovec. ROLAND: Graph Learning Framework for Dynamic Graphs. ACM Conf. on Knowledge Discovery and Data Mining 2358 - 2366 (2022)

 

 

Required knowledge

Some machine learning knowledge (including required maths; very good programming skills; some knowledge of data visualisation would be a desirable.