Projects
OPENVERSE : OPEN and co-created metaVERSe for Europe Vrije Universiteit Brussel
responsible, but also capable of restoring the technological sovereignty of EU industry in the global scenario of competition. Such a
European Metaverse would foster the following benefits: a) Higher accessibility, greater freedom and control, lower barriers to entry
and more equitable ...
A Hybrid Approach for Condition Monitoring of Machinery KU Leuven
The main maintenance approaches able to monitor complex equipment conditions for diagnostic and prognostic purposes are model-based, statistical, and Artificial Intelligence-based; to the best candidate’s knowledge, little efforts have been put into approaches’ hybridization so far. On the whole, a lack of clarity on this issue appears from the literature analysis: some authors, indeed, consider an erroneously called hybrid approach what it ...
Real-time monitoring of animal-based behavior parameters for optimal welfare-friendly captive birds KU Leuven
In this work, an efficient and relatively lightweight multimodal fusion model that considers both audio and video data is proposed to identify captive birds disease states, realize online monitoring, reduce human labor, improve work efficiency, achieve early, fast, and efficient early warning, and prevent large-scale outbreaks of captive birds diseases.
Age and sex-appropriate 3D shape analysis and geometric deep learning for children undergoing craniofacial diagnostics and surgery KU Leuven
The sole aim of many procedures in plastic and reconstructive surgery is to restore the human form but the planning and review of the outcomes of such procedures is a major challenge. 3D photography provides a radiation free, accurate and reproducible record of the craniofacial surface anatomy and has the potential to be a versatile tool for treatment planning and assessment. However, a crucial issue in the analysis of children is the change ...
Development of an innovative multiparameter sensing platform based on integrated impedance and thermal sensors for the screening of interface phenomena and changes in bulk liquid for lab-on-chip applications. Hasselt University
DUST: Designing Interpretable and Efficient Deep Unfolding Sparse Transformers for Multimodal Image Processing and Generation Vrije Universiteit Brussel
and image processing tasks. With recent developments in
autonomous driving, robotics and other smart devices, there is a
growing need for small and efficient deep learning models that are
also interpretable and whose decisions can be clearly explained to
humans. The Transformer architecture has recently gained popularity
in the computer ...
The impact of magnetic particulate matter (mPM) pollution as a contributing factor for development of childhood cancer: an in vitro and in vivo molecular and cellular analysis. Ghent University
Pediatric cancer is rare, yet it is the second leading cause of death in children. The causes of childhood cancer are not yet fully known, but it is believed that both genetic and environmental factors play a role. Recently it became clear that 10% of new diagnoses of childhood cancer can be explained by an underlying genetic predisposition. In addition, environmental factors were also investigated, including high doses of ionizing radiation, ...
Development and deployment of physics-based digital twins for the analysis of the vibro-acoustic behavior of systems of industrial complexity. KU Leuven
The vibro-acoustic performance of mechatronic systems is a vital aspect of their operation. Thus, it needs to be assessed during both the design process and the operational life of the asset. This proposal aims to develop and deploy physics-based virtual entities called “Digital Twins” (DT) for the purpose of analyzing the vibroacoustic behaviour of these systems. A DT is a virtual duplicate of a physical system, built from a fusion of ...
Towards precision health by enabling multimodal monitoring in real-life settings using uncertainty based hierarchical and time-dynamic models Ghent University
I will construct a multimodal and dynamic hierarchical sensing framework to tackle the challenges of personalized health monitoring in real-life settings. Multimodal sensing allows me to detect non-physiological symptoms by incorporating context. By fusing behavior modeling with hierarchical anomaly detection using an active learning approach, I will define the optimal moment to gather user feedback for the time dynamic models.