Adaptive automated machine learning
Can data science make data scientists redundant? I don't quite think so, but it can hopefully assist them in sometimes tedious tasks and free up their time to be more creative. Enter AutoML! Currently I am interested in the automatic development of adaptive machine learning models, the research that I have been involved in since 2012, starting from my PhD project. Together with my collaborators, Bogdan Gabrys and Damien Fay, I have published several papers in this topic, with the recent one "Automated Adaptation Strategies for Stream Learning" being published in Machine Learning Journal! Check out the code or get in touch if interested in re-implementation.
Adaptive learning for object detection/recognition
An application for adaptive learning is the computer vision, where many applications require constant update/adaptation of the model. Currently I am working on this topic with Marcin Budka and our students Joanna Stanislawek and Khairul Mottakin. I had also been awarded ACORN grant of Bournemouth University to kick-off this line of research. Stay tuned for upcoming publications.
Machine Learning for injury prediction in football
Can we understand why training injuries are happening in football? What about predicting them in advance? We are currently working on this problem together with Tim Rees, our student Aritra Majumdar and AFC Bournemouth football club. Our paper "Machine Learning for Understanding and Predicting Injuries in Soccer" has been freshly submitted to the Sports Medicine Journal.
Automated Medication Dosage
Together with collaborators from NHS and BU, I am working on estimation of medication dosage. This can automate some of the trivial tasks that doctors are charged with, and save their valuable time for more challenging issues.