On October 9th 2024 at 16:30, Esther Rodrigo Bonet will defend their PhD entitled “EXPLAINABLE AND PHYSICS-GUIDED GRAPH DEEP LEARNING FOR AIR POLLUTION MODELLING”.

Everybody is invited to attend the presentation in room I.0.02.

Air pollution has become a worldwide concern due to its negative impact on the population’s health and well-being. To mitigate its effects, it is essential to monitor pollutant concentrations across regions and time accurately. Traditional solutions rely on physics-driven approaches, leveraging particle motion equations to predict pollutants’ shifts in time. Despite being reliable and easy to interpret, they are computationally expensive and require background domain knowledge. Alternatively, recent works have shown that data-driven approaches, especially deep learning models, significantly reduce the computational expense and provide accurate predictions; yet, at the cost of massive data and storage requirements and lower interpretability.

This PhD research develops innovative air pollution monitoring solutions focusing on high accuracy, manageable complexity, and high interpretability. To this end, the research proposes various graph-based deep learning solutions focusing on two key aspects, namely, physics-guided deep learning and explainability.

First, as there exist correlations among the data points in smart city data, we propose exploiting them using graph-based deep learning techniques. Specifically, we leverage generative models that have proven efficient in data generation tasks, namely, variational graph autoencoders. The proposed models employ graph convolutional operations and data fusion techniques to leverage the graph structure and the multi-modality of the data at hand. Additionally, we design physics-guided deep-learning models that follow well-studied physical equations. By updating the graph convolution operator of graph convolutional networks to leverage the physics convection-diffusion equation, we can physically guide the learning curve of our network.

The second key point relates to explainability. Specifically, we design novel explainability techniques for interpretable graph deep modeling. We explore existing explainability algorithms, including Lasso and a layer-wise relevance propagation approach, and go beyond them to our graph-based architectures, designing efficient and specifically tailored explanation tools. Our explanation techniques are able to provide insights and visualizations based on various input data sources.

Overall, the research has produced state-of-the-art models that combine the best of both (physics-guided) graph-deep-learning-based and explainable approaches for inferring, predicting, and explaining air pollution. The developed techniques can also be applied to various applications in modeling graphs on the Internet such as in recommender systems’ applications.

Knowledge Engineering in Diagnostic Imaging – a huge project on AI in medical image analysis, the foundation of the internal ETRO ICT computer network.

The Belgian initiative icovid, which supports radiologists in the assessment of CT images of the lungs of COVID-19 patients, has grown into a multicentre European project, co-funded by the EU Horizon 2020 programme. icovid was set up in March by UZ Brussel, KU Leuven, icometrix and ETRO, an imec research group of VUB. Professor Jef Vandemeulebroucke of ETRO: “What started as a local project is now being rolled out in 800 hospitals in Europe and supported by excellent research centres all over Europe. With icolung, we can detect COVID-19 patients at an early stage and quantify the extent of lung lesions. Meanwhile, we are further improving the AI software to identify lung damage as COVID-19 even more quickly, and to determine the further care path of the patient faster and better through prognostic models.”

The icovid project was launched in March 2020 as a Belgian pro bono initiative. icometrix, which specialises in AI solutions for medical images, partnered with UZ Brussel, KU Leuven, VUB and imec to investigate how to deploy lung scans in the COVID pandemic and what AI software would be needed to do so. The AI tool icolung was born. Prof Johan de Mey, VUB-UZ Brussel: “At the time, there was insufficient testing capacity to quickly test all patients. With icolung, we wanted to use lung scans as a triage tool. By using CT and with the help of the AI analyses, we were able to trace patients with suspicious lung lesions and have them tested as a priority. During the busiest periods, everyone who entered the UZ Brussel as a patient was scanned, also as a means of preventing COVID-19 outbreaks in the hospital.”

icolung is free for all hospitals in Europe

Students who think they have had the same course before, have to fill out the form for exemption, (https://student.vub.be/en/ir#regulations-and-forms) and together with the transcripts send everything to the faculty secretary office.

Jan Cornelis – unoffcial HoD ad interim

The program is organised to accommodate your scientific background and future-oriented academic interests – developing the necessary Computer Science and Data Science skills by complementing your primary field of expertise. Above all that, we offer a wide variety of highly specialised elective courses;

The Master of Applied Computer Science provides a broad education in data science and engineering with focus on generic smart systems design, complemented with elective minors in digital health, smart cities, environmental informatics, and business intelligence. The accumulated knowledge will give rise to an ICT engineer, capable to design systems of systems and apply analytics on the heterogeneous data obtained by such systems.