PhD in Computing Science - Interactive Model-based Probabilistic Visualisations for Exploring Decisions

This studentship is linked to the DIFAI project

Applicants are invited for a fully funded PhD studentship (international fees + stipend at research council rates) in a collaborative project between the University of Glasgow and Aegean Airlines (Athens, https://en.aegeanair.com/). Aegean Airlines is interested in interactive systems to explore probabilistic models of interactions between Aegean, their customers, and their competition.

Bayesian Probabilistic models and Structural Causal models are powerful tools for Simulation Intelligence but are cumbersome to work with in practice. Exploratory queries on these models (multiverse analyses) are associated with delays that make interaction challenging, as they usually involve changes to model structure, resampling (MCMC), or retraining (Variational Inference). Pre-computing and caching results for all possible queries is infeasible due to continuous-valued causes, and an exponential number of model structures and hypothetical interventions in the number of variables. Predicting human intent - what a user might want to look at next; what are interesting aspects to focus on - can help make caching practical and thereby bridge the interaction gap, enabling interactive visualisations to be proactive. Active Inference is a promising sequential decision-making framework for simultaneous reasoning about user objectives, and selection of actions that facilitate those objectives.

This PhD will focus on investigating how aspects of active inference can be used to model human sequential interaction with probabilistic models, representing the user as an active inference agent, and to anticipate what users want to or ought to see next, representing the User Interface as active inference agent that infers and acts upon user goals. Outcomes of this project may contribute to the literature on explainable AI.

This studentship will focus on

• Embedding predictive sampling/ retraining and emulation in the UI action space to bridge the interaction latency gap
• Developing interactive visual interfaces to facilitate sequential exploration of causal models
• Producing software to manage versions of models, posterior predictive distributions and human behaviour

This PhD will be split between the University of Glasgow campus (the expectation is that the student will normally be in the lab, in person) and the Aegean Airlines offices in Athens, Greece. Students will have access to the training and academic opportunities offered at Glasgow as well as the opportunity to interact closely with Aegean Airlines' business and data teams - to learn about their business decision-making processes and machine learning production pipelines; for feedback on prototypes and participation in user studies; to work with large-scale real-world datasets and potentially run experiments in production.

The successful candidate will have a strong interest/ background in human-computer interaction and machine learning, particularly in visualisation, and/ or Bayesian probabilistic modelling.

The PhD will be supervised by Dr. Sebastian Stein and Prof. Roderick Murray-Smith at the University of Glasgow.

Candidates will be expected to hold at least a 2:1 BSc degree in Computer Science or a closely-related discipline. 

Suitable candidates will have a background or strong interest in data science and statistical modelling, particularly in Bayesian inference/probabilistic programming/causal inference, and in human-computer interaction/visualisation/interaction design.

How to Apply: 

1. Please refer to the following website for details on how to apply. http://www.gla.ac.uk/research/opportunities/howtoapplyforaresearchdegree/
2. Please email the PhD supervisor to confirm that you have submitted your application.