Data Science research theme

PhD Title: Colours-of-the-Wind (COLD): A novel framework for image-based air quality prediction using deep learning and spatial-temporal analytics

Research: My PhD research is centred on creating a novel framework incorporating image analytics as a data source for air quality analysis. The goal is to devise a method that employs computer vision to assess air quality. Building on these evaluations, I aim to develop new algorithms that integrate data from both images and monitoring station readings, enabling the use of dual data sources for more accurate air quality prediction.

Director of Studies/First Supervisor: Professor Kamran Munir
Second Supervisor: Dr Paul Matthews

PhD Title: Optimal control of aquaponic farms for food production using Internet of Things (IoT) and Artificial Intelligence (AI)

Research: My research focuses on the optimal control of aquaponics farm using IoT and AI. It aims to optimise environmental parameters such as temperature, pH, and dissolved oxygen, while forecasting energy supply and demand to enhance sustainability. By leveraging sensor data and anomaly detection, the study seeks to reduce energy footprints and ensure consistent farm performance. This work has the potential to revolutionize aquaponics farming by integrating advanced data acquisition and AI-driven insights, paving the way for more efficient and eco-friendly food production systems.

Director of Studies/First Supervisor: Professor Kamran Munir
Second Supervisor: Professor Mark Hansen 
Industry Collaboration / Supervisor: Sciflair, Dr M. Fahim Tariq

PhD Title: Autonomous and platform agnostic horizontal autoscaling of virtual machines across hybrid cloud

Research: The research focuses on extending horizontal autoscaling of virtual machines in hybrid clouds. It examines using supervised and unsupervised machine learning for time series forecasting to improve proactive autoscaling, enhancing resource and cost efficiency of autoscaling scenarios, where the current commercial public clouds provide a limited support. Also, as the other main goal, the research investigates the potential of extending the horizontal autoscaling of Virtual Machines across the hybrid cloud with the aid of a common API platform operating at the Infrastructure as a Service (IaaS) layer in cloud infrastructures.

Director of Studies/First Supervisor: Professor Kamran Munir
Second Supervisor: Professor Zaheer Khan

PhD title: Scenario-driven adaptive game design for personalized learning in programming education

Research: The PhD research introduces I-PLATO (Interoperable Personalized Learning with Adaptive Training Object), a novel framework that combines personalisation, motivation, and adaptive learning to transform programming education. By dynamically adjusting task difficulty, leveraging Game Learning Analytics, and decoupling adaptive mechanisms from specific games, I-PLATO fosters individualised, engaging, and scalable learning experiences.  The game GhostCoder demonstrates its potential to enhance learner engagement, motivation, and skill acquisition. The research further validates I-PLATO’s adaptability across diverse educational games, positioning it as a groundbreaking framework for advancing personalised and adaptive learning in programming education. 

Director of Studies/First Supervisor: Professor Kamran Munir

Second Supervisor: Dr Yasmine Rosunally

Industry Collaboration/Supervisor: SHAPE, Hong Kong 

PhD title: Automating predictive maintenance: a framework for aircraft anomaly detection and forecasting

Research: The amount of data recorded by modern aircraft is growing much faster than the supply of trained data engineers who can exploit it for predictive analytics. In response to the increasing need for predictive maintenance (PdM) in aviation - driven by the rising number of aircraft in operation - my PhD research proposes a framework that simplifies PdM model development for non-specialists. By automating parts of the PdM process, this research empowers technicians and engineers with minimal programming experience to build reliable forecasting and anomaly detection models for aircraft components. Using Airbus and synthetic datasets, this study demonstrates that AutoML can perform comparably to traditional methods, with the potential to reduce model development time and broaden the applicability of PdM solutions across the aerospace manufacturing industry.

Director of Studies/First Supervisor: Professor Kamran Munir

Second Supervisor: Dr Ahsan Ikram

Industry Collaboration/Supervisor: Airbus, UK, Dr Murad El-Bakry