Michael P. J. Camilleri

“Some people worry that artificial intelligence will make us feel inferior, but then, anybody in his right mind should have an inferiority complex every time he looks at a flower.” —Alan Kay

I am the ACRC Research Fellow (equivalent to Lecturer) at the School of Informatics (University of Edinburgh) and Associate Staff with the School of Science and Engineering (University of Dundee). I am also a CHAI Scholar (Early-Career Researcher) within the School of Engineering (University of Edinburgh). I hold a PhD in Data Science (University of Edinburgh, EPSRC Centre for Doctoral Training in Data Science), an M.Sc. in Artificial Intelligence and Robotics (University of Edinburgh), and a B.Sc. in Communications and Computer Engineering (University of Malta).

My research interests lie at the intersection of Probabilistic Machine Learning and Deep Learning applied to Clinical Data, particularly medical imaging to predict future disease onset. Throughout my career, I have worked with clinical data (SCANDAN, DECOVID, BrainIT), mouse-behaviour data (Dissertation, PhD, CDT in Data Science, University of Edinburgh), robotics (Dissertation, M.Sc. AI in Intelligent Robotics, University of Edinburgh, part of the DARPA Robotics Challenge with team HKU), astronomy (preprocessing pipeline for the Square-Kilometer Array), transport (DRT solutions at the University of Malta) and mobile gaming (RemoteFX cloud-computing, University of Malta). As a student I have also interned at CERN (with the ATLAS Experiment).

Upcoming Positions

Here I list open positions with my group or those of affiliated researchers.

Another PhD opportunity funded by the Future Medicine Program, this time with my collaborators at the College of Medicine (Miguel Bernabeu, Ting Shi) and School of Engineering (Sotos Tsaftaris) for UK (home-status) students. This project seeks to establish causal relationships between respiratory infections, vaccinations, and dementia risk by analysing retinal biomarkers, aiming to identify modifiable factors for dementia prevention and informing public health strategies to reduce neurodegenerative disease incidence. Application is through the Future Medicine Program [deadline 12/05/2026]

Research

My research interests lie at the intersection of Probabilistic Machine Learning and Deep Learning applied to Clinical Data, particularly medical imaging to predict future disease onset, but I am also motivated by general applications of computer vision, particularly to real-world problems. I am particularly intrigued by the problem of messy/un-curated data and how explicitly modelling its limitations can improve performance of models.

A full list of my publications can be found on Google Scholar

Selected Projects

A Project to classify the behaviour of group-housed mice and subsequently model their group dynamics even across cages.

Paper
Code

The Tracking and Identification Module (TIM) is a solution to track and consistently identify group-housed mice using a more challenging side-view camera setup.

Paper
Code

The Inter-Schema AdapteR is a probabilistic model for combining information from multiple annotators across different annotation schemas.

Paper
Code

VJAGG is an Android/iOS application developed to serve as an automated travel diary for transport-related research.

Paper
Code

The EXOTica library is a generic Optimisation Toolset for Robotics platforms, written in C++ with bindings for Python, aiming to provide a more streamlined process for developing algorithms for tasks such as Inverse Kinematics and Trajectory Optimisation.

Paper
Code

MPC Tools (Michael P Camilleri’s Toolbox) is a collection of python utilities that I have found useful in my data science endeavours. I hope that it will be useful to other software engineers.

Code

My Ramblings

All Posts

Reaching a Consensus in crowdsourced data using the Dawid-Skene Model

22nd Jun 202025th Jun 2020 by michaelpjcamilleri

Introduction Machine Learning and supervised machine learning in particular often requires access to human annotated data. Crowdsourcing, where the general public is given small chunks of datasets to… Read more “Reaching a Consensus in crowdsourced data using the Dawid-Skene Model”

Variational Inference for Bayesian Univariate Gaussian

24th Mar 202022nd Jun 2020 by michaelpjcamilleri

In this post, I will try to illustrate Variational Inference methods, in particular the factorised approximation approach using the Bayesian treatment for a Uni-variate Gaussian. Given that… Read more “Variational Inference for Bayesian Univariate Gaussian”

Using PCA to preserve privacy

5th Nov 201931st Mar 2020 by michaelpjcamilleri

So my Supervisor recently proposed the following problem to me as a potential way of getting the students to think about Principal Components Analysis. I write below… Read more “Using PCA to preserve privacy”