Alison Heppenstall is Professor of Geocomputation at the University of Leeds and an ESRC-Turing Fellow. She has a PhD in Artificial Intelligence, a focus of which was developing machine learning algorithms (e.g. neural networks) and AI approaches (agent-based models) that were applied to solving complex spatial problems. Most of her current research is focused on developing and adapting ML approaches to understanding social phenomena. She has particular interests in data analytics, developing approaches for detecting 'hidden' spatio-temporal patterns in 'big data', quantifying uncertainty in simulations and building more robust individual-based models through probabilistic programming and reinforcement learning.


Understanding how cities work is of increasing importance to both policy makers and practitioners. However, we currently lack the tools to be able to understand and forecast the impacts of short-term events such as road closures across the city, or answer longer-term questions about how to make our cities more sustainable and energy efficient. Different forms of data ranging from ‘social’ data (e.g. mobile phone, social media) that provide glimpses into how individuals use the city through to ‘smart’ data giving information on pollution and traffic levels are now abundant. Bringing these data together allows complex questions such as ‘what will be the impact on individual health and the local economy of making public transport free?’ to be posed. However, by adding in methods from machine learning, we can begin to both reveal the processes within cities and begin to build digital twins that allow us to simulate them. This talk will present current work on understanding and simulating cities focusing on the following questions: Can we build individual-level models that represent how humans use the city? How do we go about using AI to simulate the city in real-time? Can we quantify the uncertainty in our predictions?

Professor Alison Heppenstall

Alan Turing Institute, University of Leeds, UK